Method, apparatus, and system for providing an AI-based total life-care platform service for pets

Abstract

An apparatus according to one embodiment receives pet registration data and guardian condition data indicating the guardian's service usage conditions from a guardian's terminal, collects pet management data including health history data, biometric data, lifestyle record data, and service usage history data corresponding to the pet registration data, preprocesses the pet registration data and pet management data according to preset preprocessing criteria to generate feature data for life cycle analysis, inputs the feature data for life cycle analysis into a pre-trained AI model to calculate pet life cycle status data and care need data, generates service matching data corresponding to at least one of a well-aging service, end-of-life care service, memorial service, and healing content provision service based on the life cycle status data, care need data, and guardian condition data, provides service provision information corresponding to the service matching data to the guardian's terminal, generates feedback data for AI model training based on response data to the service provision information received from the guardian's terminal and usage result data corresponding to the service provision information, and stores the feedback data for AI model training in a pet life cycle learning database.

Description

Technology Field

[0001] The following embodiments relate to technology providing a method, device, and system for providing an AI-based total life care platform service for pets. Background Technology

[0003] With the recent increase in pet owners recognizing their companion animals as family members, there is a growing demand for support services related to pet health management, daily living, funerals, memorial services, and owner healing. In particular, as the average lifespan of pets increases, owners desire continuous, appropriate care tailored to their pets' age, health status, and changing living conditions, rather than simple one-time maintenance services.

[0004] However, conventional pet-related services—such as health checkups, test kits, food or supplies, funerals, memorial spaces, and healing programs—are often provided as separate, independent services. Consequently, pet owners must personally select services suitable for their pet's current condition, while service providers find it difficult to continuously reflect the pet's health status, living conditions, and service utilization results.

[0005] Furthermore, conventional pet funeral or memorial services often focus on the periods before and after the pet's death. Since this approach fails to organically connect information regarding pre-death care with support for post-death memorial or healing content, it has limitations in providing management services that consider the pet's entire life cycle.

[0006] Meanwhile, a pet's condition can vary depending on various factors, such as species, breed, age, weight, medical history, activity level, intake, bowel movements, sleep patterns, and lifestyle habits. Nevertheless, conventional service recommendation methods are often based on the product type selected by the owner, simple age classifications, or pre-determined service categories, making it difficult to adequately reflect actual changes in the pet's condition and the owner's usage conditions.

[0007] Furthermore, there is a problem in that usage results after the initial service provision are not systematically reflected in subsequent service provision. For instance, there is a lack of a structure to learn how a pet's health status or lifestyle patterns changed after using a specific care service, or whether the service was suitable for other pets in similar conditions, and to utilize this knowledge for the provision of subsequent services.

[0008] Some conventional pet care services provide structures that assess health status or recommend health management services based on pet registration information, growth management data, life log data, or medical history questionnaires. However, these methods often limit service decisions to status data prior to service provision or information entered by the owner, and lack a structure that systematically converts changes in the pet's condition, test results, or lifestyle patterns after service use into subsequent AI model training data.

[0009] Furthermore, while conventional health-focused services may be useful for assessing a pet's current health status or providing health management services, they have limitations in connecting pre-life care, end-of-life care, post-burial memorial services, and healing content for guardians into a single life cycle flow. In particular, there is a need for technology capable of generating service provision information based on a pet's life cycle status and care needs, creating feedback data for learning regarding changes in condition before and after service use, and reflecting life cycle-specific measurement and management information—based on similar populations and statistical standards—in subsequent service matching.

[0010] Therefore, technology is required to manage the entire process—from pre-life care to end-of-life care and support for memorial and healing content—as a single continuous service system, and to provide appropriate services by analyzing changes in the pet's condition and the guardian's usage conditions based on data. Furthermore, AI-based total life care platform technology for pets is required that can continuously improve the accuracy of service provision for each pet by analyzing and utilizing post-service usage results for learning. Prior art literature

[0012] Republic of Korea Registered Patent No. 10-2422223 (Published July 15, 2022) Republic of Korea Registered Patent No. 10-2706826 (Published September 20, 2024) Republic of Korea Registered Patent No. 10-2894952 (Published December 4, 2025) Republic of Korea Registered Patent No. 10-2531855 (Published June 1, 2023) The problem to be solved

[0013] The embodiments aim to provide a method for generating service provision information tailored to the life cycle status and care needs of pets.

[0014] The embodiments aim to provide a method for generating feedback data for training an AI model based on changes in state before and after service use.

[0015] The embodiments aim to provide a method for selecting key measurement items by life cycle and generating corresponding measurement management information.

[0016] The embodiments aim to provide a method for generating end-of-life management information tailored to the end-of-life condition of a pet and the conditions of its guardian.

[0017] The embodiments aim to provide a method for generating post-care information based on usage history following funeral or memorial services.

[0018] The objectives of the present invention are not limited to those mentioned above, and other unmentioned objectives will be clearly understood from the description below. means of solving the problem

[0020] According to one embodiment, a method for providing an AI-based total life care platform service for pets comprises: receiving pet registration data and guardian condition data indicating the guardian's service usage conditions from a guardian's terminal; collecting pet management data including health history data, biometric data, lifestyle record data, and service usage history data corresponding to the pet registration data; preprocessing the pet registration data and the pet management data according to preset preprocessing criteria to generate feature data for life cycle analysis; inputting the feature data for life cycle analysis into a pre-trained AI model to calculate pet life cycle status data and care need data; generating service matching data corresponding to at least one of a well-aging service, an end-of-life care service, a memorial service, and a healing content provision service based on the life cycle status data, the care need data, and the guardian condition data; providing service provision information corresponding to the service matching data to the guardian's terminal; and generating feedback data for AI model training based on response data to the service provision information received from the guardian's terminal and usage result data corresponding to the service provision information. The method may include the step of storing feedback data for training the AI ​​model in a pet life cycle learning database, wherein the AI ​​model is trained based on the pet life cycle learning database and is configured to generate the life cycle status data and the care need data as output data using the feature data for life cycle analysis as input data.

[0021] The step of generating feedback data for training the AI ​​model comprises: generating service usage characteristic data by extracting service identification information, service selection status, service usage time, and pet condition change information from the response data and usage result data; matching the pet condition change information after service usage with the pet registration data, health history data, biometric data, and lifestyle record data based on a pre-assigned pet identifier; sorting the matching results in the order of collection time to generate pet-specific life cycle history data; excluding missing data, abnormal data, and data that does not satisfy observation period criteria from the life cycle history data according to pre-set exclusion criteria; dividing the life cycle history data into reference interval data prior to the service usage time and observation interval data after the service usage time; calculating the amount of change in health status, the amount of change in test results, and the amount of change in lifestyle patterns by comparing the reference interval data and the observation interval data; and assigning life cycle labels according to pre-set life cycle label assignment criteria based on the pet registration data, health history data, biometric data, lifestyle record data, the amount of change in health status, the amount of change in test results, and the amount of change in lifestyle patterns. Steps for assigning a care result label by comparing the change in health status, the change in test results, and the change in lifestyle patterns with pre-set care effect judgment criteria; generating training candidate data by matching the service usage characteristic data, the life cycle label, and the care result label with each other; classifying the training candidate data by combinations of the life cycle label and the care result label; and when the difference in data distribution by combination of the life cycle label and the care result label satisfies a pre-set imbalance criterion,It may include a step of correcting the sample ratio of the data for each combination, and a step of generating feedback data for training the AI ​​model based on the training candidate data reflecting the result of correcting the sample ratio.

[0022] After the step of calculating the life cycle status data and the care need data, and before the step of generating the service matching data, the method may further include the step of generating life cycle-specific measurement management information, and the step of setting improvement service candidate data included in the life cycle-specific measurement management information as service candidate data used to generate the service matching data according to preset candidate setting criteria.

[0023] The step of generating the above-mentioned life cycle measurement management information comprises: a step of extracting similar population data by comparing the above-mentioned pet registration data, the above-mentioned life cycle status data, and the above-mentioned health history data with individual history data stored in the above-mentioned pet life cycle learning database according to preset criteria for selecting similar individuals; a step of calculating impact data by measurement item by applying preset impact analysis criteria to at least one of the activity level management need value, intake level management need value, bowel state management need value, sleep level management need value, and weight change management need value included in the above-mentioned care need data; a step of extracting items where the impact data by measurement item is greater than or equal to the preset impact criteria as core factor data; a step of generating precise measurement condition data including measurement period, measurement method, and measurement precision conditions corresponding to the above-mentioned core factor data; a step of collecting core factor measurement data corresponding to the above-mentioned core factor data according to the above-mentioned precise measurement condition data; a step of calculating measurement validity data by comparing the above-mentioned core factor measurement data with statistical standard data of a pet group belonging to the same species and the same life cycle; a step of generating measurement deviation data and improvement need item data by comparing the above-mentioned core factor measurement data with corresponding measurement data included in the above-mentioned similar population data; and the above-mentioned measurement validity data, The method may include the step of generating improvement service candidate data based on the measurement deviation data and the improvement-needing item data, and the step of generating lifecycle-specific measurement management information including the precision measurement condition data, the measurement validity data, the measurement deviation data, the improvement-needing item data, and the improvement service candidate data.

[0024] After the step of generating the above service matching data, the method may further include the step of generating end-of-life care information and the step of including the above end-of-life care information in the above service provision information.

[0025] The step of generating the above-mentioned end-of-life management information comprises: a step of extracting the level of need for end-of-life management from the care need data; a step of extracting pet characteristic data including species, breed, weight, and age from the pet registration data; a step of extracting health restriction data including disease status information and movement restriction information from the health history data and the life record data; a step of generating end-of-life management stage data by comparing the life cycle status data, the level of need for end-of-life management, the pet characteristic data, and the health restriction data with preset end-of-life stage judgment criteria; a step of extracting target service data corresponding to the end-of-life management service type from the service matching data; a step of extracting provision condition data including target pet conditions, preparation period, available regions, provision method conditions, and information required for guardian input from the target service data; a step of extracting guardian detailed condition data including regional conditions, pet burial method conditions, pet transportation conditions, and memorial content conditions from the guardian condition data; a step of calculating pet suitability data by comparing the above-mentioned pet characteristic data and the above-mentioned health restriction data with the target pet conditions; a step of calculating condition suitability data by comparing the above-mentioned guardian detailed condition data with the above-mentioned provision condition data; and the above-mentioned pet A step of selecting available service data among the target service data based on suitability data and the condition suitability data; a step of calculating the execution priority of the available service data based on the end-of-life management stage data and the level of need for end-of-life management; a step of setting the scheduled execution time for each service based on the execution priority and the preparation period; a step of setting guardian input request conditions and guardian notification conditions based on the guardian input required information and the scheduled execution time for each service.The method includes the step of generating end-of-life management information including the end-of-life management stage data, the available service data, the execution priority, the scheduled execution time for each service, the guardian input request condition, and the guardian notification condition, wherein the end-of-life management service type may include at least one of a pet funeral preparation type, a pet transportation support type, a pet post-mortem care type, a memorial content creation type, and a memorial space usage type.

[0026] The method further includes the step of generating post-care information based on usage result data corresponding to the above service provision information, and the step of providing the above post-care information to the guardian's terminal, wherein the step of generating the post-care information comprises: the step of extracting pet funeral completion information and memorial service initiation information from the above usage result data; the step of collecting post-care usage history data including at least one of electronic memorial space access history, memorial content viewing history, memorial message input history, guestbook input history, and healing content usage history from the guardian's terminal when at least one of the pet funeral completion information and the memorial service initiation information satisfies a preset post-care initiation criterion; the step of generating post-care usage characteristic data including access frequency, usage duration, number of content usages, and whether healing content is used from the above post-care usage history data; the step of extracting the guardian's memorial content preference condition, healing content preference condition, and community linkage allowance condition from the above response data; the step of extracting post-care service types including electronic memorial space type, virtual memorial content type, art-based healing content type, meditation-based healing content type, counseling linkage type, and community linkage type from pre-stored post-care candidate data; and the above post-care usage characteristic data, the A step of calculating memorial usage duration data and healing content provision necessity data based on memorial content preference conditions, healing content preference conditions, and community linkage permission conditions; a step of calculating memorial service suitability data by comparing the memorial usage duration data and memorial content preference conditions with the electronic memorial space type and the virtual memorial content type; and a step of comparing the healing content provision necessity data, healing content preference conditions, and community linkage permission conditions with the art-based healing content type, the meditation-based healing content type,The method may include the steps of: calculating healing support suitability data by comparing it with the counseling linkage type and the community linkage type; calculating the priority of provision for each post-care service based on the memorial service suitability data and the healing support suitability data; calculating the amount of change over time intervals of the post-care usage characteristic data and setting the memorial content provision cycle and healing content notification conditions based on the priority of provision for each post-care service and the amount of change over time intervals; and generating the post-care information including the memorial service suitability data, the healing support suitability data, the priority of provision for each post-care service, the memorial content provision cycle, and the healing content notification conditions.

[0027] The step of generating the above service matching data comprises: extracting health management need level, nutritional management need level, activity management need level, screening management need level, end-of-life care need level, and aftercare need level from the above care needs data; extracting budget condition data and usage restriction conditions from the above guardian condition data; extracting service identification information, service type, cost condition, provision timing condition, and provision exclusion condition from pre-stored service candidate data; calculating management suitability data by comparing the above care needs data with the above service type; calculating time timing suitability data by comparing the above life cycle status data with the above provision timing condition; calculating provision possibility data by comparing the above budget condition data and the above usage restriction condition with the above cost condition and the above provision exclusion condition; calculating a matching score for each service candidate based on the above management suitability data, the above time timing suitability data, and the above provision possibility data; selecting service candidates whose matching score for each service candidate is greater than or equal to a pre-set matching criterion; classifying the selected service candidates into essential service candidates and optional service candidates according to the above care needs data and pre-set provision criteria by management area; and within the range of the above budget condition data, the above essential service candidates The method may include the steps of: generating service combination data by first arranging the selected service candidates in order of highest matching score for each service candidate; setting the service provision order and service notification conditions included in the service combination data based on the life cycle status data and the care needs data; and generating service matching data including the service combination data, the matching score for each service candidate, the service provision order, and the service notification conditions.

[0028] A device according to one embodiment may be combined with hardware and controlled by a computer program stored on a medium to execute the method of any one of the methods described above. Effects of the invention

[0030] The embodiments can provide a method for generating appropriate service provision information based on the life cycle status and care needs of the pet.

[0031] The embodiments can provide a method to improve service matching accuracy by reflecting state changes before and after service use in the training of an AI model.

[0032] The embodiments can provide a method for precisely deriving necessary management items by generating measurement management information on key factors for each life cycle.

[0033] The embodiments can provide a method for systematically generating end-of-life management information by reflecting the end-of-life condition of the pet and the guardian's conditions.

[0034] The embodiments may provide a method for continuously providing memorial and healing support based on usage history following funeral or memorial services.

[0035] Meanwhile, the effects according to the embodiments are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art from the description below. Brief explanation of the drawing

[0037] FIG. 1 is a schematic diagram showing the configuration of a system according to one embodiment. FIG. 2 is a flowchart illustrating the process of providing an AI-based total life care platform service for pets according to one embodiment. FIG. 3 is a flowchart illustrating the process of generating feedback data for training an AI model according to one embodiment. FIG. 4 is a flowchart illustrating the process of generating life cycle measurement management information according to one embodiment. FIG. 5 is a flowchart illustrating the process of generating end-of-life care information according to one embodiment. FIG. 6 is a flowchart illustrating the process of generating post-management information according to one embodiment. FIG. 7 is a flowchart illustrating the process of generating service matching data according to one embodiment. FIG. 8 is an example diagram of the configuration of a device according to one embodiment. Specific details for implementing the invention

[0038] Hereinafter, embodiments are described in detail with reference to the attached drawings. However, various modifications may be made to the embodiments, and thus the scope of the patent application is not limited or restricted by these embodiments. It should be understood that all modifications, equivalents, and substitutions to the embodiments are included within the scope of the rights.

[0039] Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be modified and implemented in various forms. Accordingly, the embodiments are not limited to the specific disclosed forms, and the scope of this specification includes modifications, equivalents, or substitutions that fall within the technical concept.

[0040] Terms such as "first" or "second" may be used to describe various components, but these terms should be interpreted solely for the purpose of distinguishing one component from another. For example, the first component may be named the second component, and similarly, the second component may be named the first component.

[0041] When it is stated that a component is "connected" to another component, it should be understood that it may be directly connected to or coupled with that other component, or that there may be other components in between.

[0042] The terms used in the embodiments are for illustrative purposes only and should not be interpreted as intended to be limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as "comprising" or "having" are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0043] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the embodiments pertain. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application.

[0044] In addition, when describing with reference to the attached drawings, identical components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted. In describing the embodiments, if it is determined that a detailed description of related prior art could unnecessarily obscure the essence of the embodiments, such detailed description is omitted.

[0045] The embodiments can be implemented in various forms of products, such as personal computers, laptop computers, tablet computers, smartphones, smart home appliances, intelligent automobiles, kiosks, and wearable devices.

[0046] FIG. 1 is a schematic diagram showing the configuration of a system according to one embodiment.

[0047] Referring to FIG. 1, a system according to one embodiment may include a guardian's terminal (100) and a device (200) that can communicate with each other through a communication network.

[0048] First, the communication network may be composed of a wired communication network, a wireless communication network, or a combination thereof, and may support data transmission and reception between the guardian's terminal (100) and the device (200). The communication network may include, but is not limited to, a mobile communication network, an internet network, a short-range wireless communication network, or a combination thereof.

[0049] The guardian's terminal (100) is a user terminal used by a guardian who manages a pet. The guardian can input or modify pet registration data and guardian condition data through the guardian's terminal (100).

[0050] The guardian's terminal (100) may be implemented as at least one of a smartphone, tablet computer, laptop computer, desktop computer, or wearable device, but is not limited thereto. The guardian's terminal (100) may communicate with the device (200) through an application, a web browser, or a platform access screen.

[0051] Pet registration data may include at least one of pet identification information, species, breed, age, weight, gender, and guardian identification information. Guardian condition data may include at least one of the guardian's budget condition, available area, preferred service type, notification reception condition, burial method condition, memorial content condition, and healing content condition.

[0052] The guardian's terminal (100) can collect data based on records generated during the guardian's direct input, linkage with an external device, or the process of using the service. For example, the guardian's terminal (100) can transmit at least one of the pet's activity level, intake amount, defecation status, sleep status, weight change, test kit results, hospital usage records, and service usage records to the device (200).

[0053] The guardian's terminal (100) can receive service provision information corresponding to service matching data from the device (200) and display it to the guardian. The service provision information may include guidance information, provision time information, notification information, or usage condition information regarding at least one of a well-aging service, end-of-life care service, memorial service, and healing content provision service.

[0054] The guardian's terminal (100) can transmit response data generated during the process of the guardian viewing, selecting, rejecting, reserving, using, or evaluating service provision information to the device (200). Additionally, the guardian's terminal (100) can transmit usage result data or post-use history data after the service is used to the device (200).

[0055] The guardian's terminal (100) is not limited to a configuration that directly performs analysis of the AI ​​model or service matching operations. The generation of life cycle status data, care need data, service matching data, and feedback data for AI model training using the AI ​​model can be performed by the device (200), and the guardian's terminal (100) can function as an interface for transmitting and receiving data with the device (200).

[0056] The device (200) is a computing device that performs a method of providing AI-based total life care platform services for pets. The device (200) can communicate with the guardian's terminal (100) to receive pet registration data, guardian condition data, pet management data, response data, and usage result data.

[0057] The device (200) may be implemented as a server, a cloud server, a platform operation server, a data processing server, or a combination thereof. The device (200) may be implemented as a single physical device, or it may be implemented in a form in which multiple servers are interconnected to perform data collection, data storage, AI model processing, service matching, and information provision functions, respectively.

[0058] The device (200) can generate feature data for life cycle analysis by processing pet registration data and pet management data according to preset preprocessing criteria. Additionally, the device (200) can input the feature data for life cycle analysis into an AI model to produce life cycle status data and care need data.

[0059] The device (200) can generate service matching data based on life cycle status data, care need data, and guardian condition data, and provide service provision information corresponding to the service matching data to the guardian's terminal (100).

[0060] The device (200) can generate feedback data for AI model training based on response data and usage result data received from the guardian's terminal (100), and store the generated feedback data for AI model training in a pet life cycle learning database.

[0061] In one embodiment, the device (200) may further include an encryption processing unit and a ledger management unit for customer information protection and verification of the integrity of prepaid deposit asset management information. The encryption processing unit may encrypt at least one of guardian identification information, pet identification information, health history data, biometric data, lifestyle record data, service usage history data, response data, usage result data, feedback data for AI model training, reward point data, and prepaid deposit asset management data according to a preset encryption standard.

[0062] The ledger management unit is not limited to directly storing the original encrypted data or the original personal information text on the blockchain network. The ledger management unit may record at least one of a data hash value, transaction identification information, deposit history identification information, change history information, access history information, and integrity verification value on the blockchain network or distributed ledger network.

[0063] Prepaid deposit asset management data may refer to management data related to amounts prepaid or deposited by a guardian to use subscription services, well-aging services, end-of-life care services, funeral services, memorial services, healing content provision services, or pet supply purchase services. Prepaid deposit asset management data may include at least one of deposit payment history, deposit usage history, balance information, refund history, deduction history by service, deposit asset identification information, and deposit status information.

[0064] The device (200) is not limited to a device that simply stores data input or collected from the guardian's terminal (100). The device (200) can function as a platform processing entity that performs analysis of the pet's life cycle status, calculation of care needs, service matching, generation of life cycle-specific measurement management information, generation of end-of-life care information, generation of post-care information, and generation of feedback data for AI model training.

[0065] Meanwhile, for convenience of explanation, FIG. 1 and the following description are based on one guardian terminal (100), but the number of guardian terminals (100) may vary depending on the embodiment. Within the range allowed by the processing capacity of the device (200), the number of guardian terminals (100) linked with the device (200) is not particularly limited.

[0066] Additionally, although not illustrated in FIG. 1, the device (200) may be connected to communicate with one or more affiliated animal hospital servers, affiliated care facility servers, external messaging servers, external messenger servers, SNS servers, blockchain networks, or distributed ledger networks.

[0067] In the present invention, Artificial Intelligence (AI) refers to a technology that imitates human learning ability, reasoning ability, and perceptual ability, and implements them on a computer, and may include concepts such as machine learning and symbolic logic. Machine Learning (ML) is an algorithmic technology that classifies or learns the characteristics of input data on its own. AI technology can analyze input data as a machine learning algorithm, learn from the results of the analysis, and make judgments or predictions based on the results of the learning. Furthermore, technologies that mimic the functions of the human brain, such as cognition and judgment, by utilizing machine learning algorithms can also be understood as falling within the category of AI. For example, technological fields such as linguistic understanding, visual understanding, reasoning / prediction, knowledge representation, and motion control may be included.

[0068] Machine learning can refer to the process of training neural network models using experience in processing data. It implies that through machine learning, computer software improves its own data processing capabilities. A neural network model is constructed by modeling the correlations between data, and these correlations can be expressed by multiple parameters. A neural network model extracts and analyzes features from given data to derive correlations between them; machine learning can be defined as the process of optimizing the model's parameters by repeating this process. For example, a neural network model can learn the mapping (correlation) between inputs and outputs for data given as input-output pairs. Alternatively, even when only input data is provided, a neural network model can derive regularities between the given data and learn those relationships.

[0069] An artificial intelligence learning model or neural network model can be designed to implement the structure of the human brain on a computer and may include multiple network nodes that have weights and simulate neurons of a human neural network. The multiple network nodes may have interconnected relationships by simulating the synaptic activity of neurons, where neurons exchange signals through synapses. In an artificial intelligence learning model, multiple network nodes may be located in layers of different depths and exchange data according to convolutional connections. The artificial intelligence learning model may be, for example, an Artificial Neural Network (ANN) or a Convolutional Neural Network (CNN). As an embodiment, the artificial intelligence learning model may be machine learned according to methods such as supervised learning, unsupervised learning, and reinforcement learning. Machine learning algorithms for performing machine learning may include Decision Tree, Bayesian Network, Support Vector Machine, Artificial Neural Network, Ada-boost, Perceptron, Genetic Programming, and Clustering.

[0070] Among these, CNNs are a type of multilayer perceptron designed to use minimal preprocessing. CNNs consist of one or more convolutional layers and standard artificial neural network layers stacked on top, additionally utilizing weights and pooling layers. Thanks to this structure, CNNs can fully utilize two-dimensional input data. Compared to other deep learning architectures, CNNs demonstrate good performance in both image and audio fields. CNNs can also be trained using standard backpropagation. CNNs have the advantage of being easier to train than other feedforward artificial neural network techniques and using a small number of parameters.

[0071] Convolutional networks are neural networks comprising sets of nodes with bounded parameters. Many computer vision tasks have been significantly improved, driven by the increased size of available training data and the availability of computational power, combined with algorithmic advancements such as discriminative linear units and dropout training. In the case of massive datasets, such as those available for many tasks today, outfitting is not critical, and increasing the network size improves test accuracy. Optimal utilization of computing resources becomes a limiting factor. To address this, distributed, scalable implementations of deep neural networks can be employed.

[0072] FIG. 2 is a flowchart illustrating the process of providing an AI-based total life care platform service for pets according to one embodiment.

[0073] Referring to FIG. 2, first, in step S201, the device (200) can receive pet registration data and guardian condition data indicating the guardian's service usage conditions from the guardian's terminal (100).

[0074] Pet registration data is basic data used by the device (200) to identify a pet under management within the platform and to determine the basic status of the pet. Pet registration data may include at least one of pet identification information, name, species, breed, date of birth, age, gender, weight, neutering status, guardian identification information, and registration date.

[0075] Pet identification information is information for distinguishing each pet when a single guardian registers multiple pets. The pet identification information may consist of at least one of a unique identifier generated by the device (200), a registration number entered from the guardian's terminal (100), microchip identification information, or a pet management number linked to platform member information.

[0076] Guardian condition data is data representing conditions that a guardian wishes to be reflected when using a total life care platform service for pets. Guardian condition data may include at least one of budget conditions, available regions, available hours, preferred service types, notification reception conditions, service provision method conditions, funeral method conditions, transportation conditions, memorial content conditions, and healing content conditions.

[0077] When the device (200) receives a request for data input completion, a membership registration request, a pet registration request, or a service condition setting request from the guardian's terminal (100), it can receive the input values ​​included in the request as pet registration data and guardian condition data. The device (200) can check whether required items have been entered, the data format, the numerical range, and whether there is duplicate registration for the received data.

[0078] For example, the device (200) can receive pet registration data from the guardian's terminal (100) in which the pet's species is dog, breed is Maltese, age is 12 years old, weight is 4.2 kg, gender is female, and neutering status is completed. Additionally, the device (200) can receive guardian condition data including a monthly usage budget, available areas, preference for health checkup services, intention to use end-of-life care services, funeral method conditions, reception of memorial content, and notification reception cycle.

[0079] The device (200) can store the received pet registration data and guardian condition data by correlating them with pet identification information or guardian identification information. By managing the data by correlating it with identification information in this way, the device (200) can process the data and guardian conditions for each pet separately, even if the same guardian has registered multiple pets.

[0080] Through this, the device (200) can obtain the basic characteristics of the pet and the guardian's service usage conditions as initial data. Accordingly, the device (200) can subsequently accurately link the pet's health history data, biometric data, lifestyle record data, and service usage history data with the pet, and can perform the generation of service matching data that reflects the guardian's actual usage conditions.

[0081] In step S202, the device (200) can collect pet management data including health history data, biometric data, lifestyle record data, and service usage history data corresponding to pet registration data.

[0082] Pet management data is management target data collected by the device (200) to check the health status, physical condition, living condition, and service usage status of a registered pet. Pet management data can be stored in correspondence with pet identification information included in pet registration data.

[0083] Health history data is data representing historical information related to the health status of a pet. For example, health history data may include at least one of health check-up history, medical treatment history, disease history, vaccination history, medication history, test result information, and surgery history.

[0084] Biometric data is data indicating the physical or physiological state of a pet. For example, biometric data may include at least one of body weight, body temperature, heart rate, respiratory rate, activity level, sleep amount, intake, bowel movements, urination, and test results.

[0085] Life record data is data that includes daily life-related data generated during the process of a guardian managing a pet, as well as test-related records verified by the guardian. For example, life record data may include at least one of walking records, feeding records, sleep records, defecation records, urination records, urine test results, hair test results, test kit results, play records, abnormal behavior records, records with attached photos or videos, and life notes entered by the guardian.

[0086] Service usage history data is data representing the usage history of services provided to a pet or its guardian through the platform or through external service providers linked to the platform. For example, service usage history data may include at least one of the following: usage history of well-aging services, usage history of health check-up related services, usage history of pet daycare, usage history of hotel, usage history of care facilities, usage history of boarding services, usage history of end-of-life care services, usage history of memorial services, and usage history of healing content provision services.

[0087] The device (200) can collect pet management data through direct input by the guardian via the guardian's terminal (100), measurements from an external device linked to the guardian's terminal (100), data linked with an external service server, or service processing records within the platform. For example, the device (200) can collect records of defecation status entered into the guardian's terminal (100), activity levels measured by an external device, health checkup results provided by a hospital system, or service reservation history generated on the platform.

[0088] In one embodiment, the device (200) may provide a daily record input screen for a pet to a guardian's terminal (100) through an application, a web browser, or a platform access screen. The daily record input screen may include input items for receiving at least one of a defecation record, a urination record, a walking record, a feeding record, a sleep record, a play record, an abnormal behavior record, a urine test result, a hair test result, a test kit result, and a daily memo.

[0089] The guardian's terminal (100) can receive life record data in at least one of the following methods: guardian's selection input, numerical input, text input, photo attachment, video attachment, or test result image attachment, and the device (200) can receive the life record data input from the guardian's terminal (100). For example, the guardian can input the number of bowel movements, bowel movement status, walking time, walking distance, urine test result, hair test result, or test kit reading result through the guardian's terminal (100).

[0090] The device (200) can store lifestyle record data received from the guardian's terminal (100) by correlating it with pet identification information, guardian identification information, input time, and input items. Additionally, the device (200) can store items that can be used to determine physical condition, such as urine test results, hair test results, or test kit results, by classifying them as lifestyle record data or biometric data.

[0091] The device (200) can provide information requesting the input of daily records according to record request conditions pre-set on the guardian's terminal (100). For example, the device (200) can provide an input request notification to the guardian's terminal (100) if a record of bowel movements has not been entered for a certain period, if a record of walks is missing, or if the scheduled date for entering urine test or hair test results has arrived.

[0092] According to this configuration, the device (200) can continuously collect daily records regarding the pet's defecation, walks, urination, test results, and living conditions through the guardian's terminal (100). Accordingly, the device (200) can reflect the living records directly observed by the guardian into the pet management data and increase the continuity of the living-related data used to calculate life cycle status data and care need data.

[0093] In one embodiment, when the guardian's consent data is confirmed, the device (200) may receive at least one of a health checkup record, vaccination record, deworming administration record, medication record, test result record, medical treatment record, surgery record, and veterinarian's opinion information corresponding to a pet identifier from one or more affiliated animal hospital servers. The affiliated animal hospital servers may include at least one of a medical record management server, a hospital reservation server, a test result management server, or an external medical record linkage server operated by the animal hospital.

[0094] The guardian's consent data is data indicating that the guardian has allowed data linkage with the affiliated animal hospital server through an application, web browser, or platform access screen installed on the guardian's terminal (100). The guardian's consent data may include at least one of the following: information on the hospital subject to consent, items of data subject to consent, a period for allowing data transmission, a scope of data utilization, and whether consent has been withdrawn.

[0095] The device (200) can store data received from an affiliated animal hospital server by correlating it with a pet identifier, guardian identification information, hospital identification information, and the time of collection according to a preset medical record linkage standard. The medical record linkage standard may include at least one of a data item name conversion standard, a date format conversion standard, an examination item code mapping standard, a duplicate record removal standard, and a missing item verification standard.

[0096] For example, if a guardian agrees to record linkage with a specific affiliated animal hospital through the guardian's terminal (100), the device (200) can automatically receive records of the completion of the pet's health checkup, vaccination date, deworming date, blood test results, urine test results, imaging test results, or surgery completion records from the server of the affiliated animal hospital.

[0097] The device (200) can classify the received records into health history data or biometric data and store them in a pet life cycle learning database. Additionally, the device (200) can provide all or part of the health history data or biometric data received from the affiliated animal hospital server so that the guardian can view it through an application, web browser, or platform access screen installed on the guardian's terminal (100).

[0098] According to this configuration, the device (200) does not rely solely on health information directly entered into the guardian's terminal (100), but can automatically collect records related to examinations, tests, medication, and surgery generated from an affiliated animal hospital server with the guardian's consent. Accordingly, the device (200) can reduce the omission of health history data of the pet and increase the accuracy and continuity of health-related data used to calculate life cycle status data and care need data.

[0099] In one embodiment, when the guardian's consent data is confirmed, the device (200) may receive at least one of an attendance record, a pickup record, a care usage record, a walking record, a play record, a feeding record, a watering record, a defecation record, a urination record, a nap record, an activity level record, an abnormal behavior record, a record with attached photos or videos, and a care staff observation record corresponding to a pet identifier from one or more affiliated care facility servers. The affiliated care facility servers may include at least one of a record management server, a reservation management server, or an external care record linkage server operated by a pet kindergarten, a pet hotel, a care center, a boarding facility, or a provider of pet activity programs.

[0100] In this case, the guardian's consent data may be data indicating that the guardian has allowed data linkage with the affiliated care facility server through an application, web browser, or platform access screen installed on the guardian's terminal (100). The guardian's consent data may include at least one of the following: information on the care facility subject to consent, items of data subject to consent, a period for allowing data transmission, a scope of data utilization, and whether consent has been withdrawn.

[0101] The device (200) can store data received from an affiliated care facility server by correlating it with a pet identifier, guardian identification information, care facility identification information, and the time of collection according to a preset care record linkage standard. The care record linkage standard may include at least one of a data item name conversion standard, a date format conversion standard, an activity type mapping standard, a duplicate record removal standard, a photo or video attachment processing standard, and a missing item verification standard.

[0102] For example, if a guardian agrees to record linkage with a specific affiliated pet kindergarten through the guardian's terminal (100), the device (200) can automatically receive the pet's arrival time, departure time, play activity details, walking time, feeding status, defecation status, nap time, abnormal behavior, or observation notes from the caregiver from the server of the affiliated care facility. The device (200) can classify the received records into life record data or service usage history data and store them in a pet life cycle learning database.

[0103] Additionally, the device (200) may provide all or part of the life record data or service usage history data received from the affiliated care facility server so that the guardian can view it through an application, web browser, or platform access screen installed on the guardian's terminal (100).

[0104] According to this configuration, the device (200) does not rely solely on daily records directly entered by the guardian, but can automatically collect daily management records generated at affiliated pet kindergartens or care facilities with the guardian's consent. Accordingly, the device (200) can reflect information on the living conditions, activity conditions, and abnormal behaviors of the pet during the time when the pet is separated from the guardian in the pet management data, and can increase the continuity of life-related data used to calculate life cycle status data and care need data.

[0105] In one embodiment, when life record data is registered through the guardian's terminal (100) or when pet management data is automatically transmitted from an affiliated animal hospital server or an affiliated care facility server according to the guardian's consent data, the device (200) can generate reward point data according to a pre-set reward standard in response to the results of collecting pet management data.

[0106] Reward point data may refer to reward data within the platform granted to a guardian in response to the guardian entering records related to the pet or allowing data synchronization with an external server.

[0107] The compensation criteria may include at least one of data input items, data input cycle, consent to automatic data transmission, data items subject to automatic transmission, data validity, duplicate input, scope of consent to data use, and duration of data collection.

[0108] For example, the device (200) can generate reward point data according to the type and input cycle of the input item when defecation records, walking records, urine test results, or hair test results are entered through the guardian's terminal (100). Additionally, the device (200) can generate reward point data according to the data items subject to automatic transmission or the transmission maintenance period when the guardian has agreed to automatic data transmission with an affiliated animal hospital server or an affiliated care facility server.

[0109] When generating reward point data, the device (200) can store pet identification information, guardian identification information, reason for reward occurrence, time of reward occurrence, quantity of reward points, and reward status information in correspondence with each other. The reward status information may include at least one status value among scheduled payment, payment completed, payment suspended, use completed, or expired.

[0110] The device (200) may restrict or withhold the generation of reward point data if the collected data does not satisfy preset validity criteria. For example, if the same defecation record is entered duplicately, if required input items are missing, if the inspection result image is not identified, or if the consent for automatic transmission is withdrawn, the device (200) may not generate reward point data or set the payment to a withheld state.

[0111] According to this configuration, the device (200) can increase the frequency and continuity of pet management data collection by generating reward point data corresponding to the guardian's direct record input and automatic record transmission through the affiliate server. Accordingly, the device (200) can more continuously secure pet management data necessary for calculating life cycle status data and care need data.

[0112] The device (200) can correspond the collection time, data source, and pet identification information together with the collected pet management data. Through this, the device (200) can manage health history data, biometric data, lifestyle record data, and service usage history data for each pet separately, even when the same guardian has registered multiple pets.

[0113] As a result, the device (200) can secure not only the initial registration information of the pet but also subsequent changes in health status, biological status, living status, and service usage status. Accordingly, the device (200) can reliably prepare data necessary for subsequent analysis by reflecting the actual management status of the pet.

[0114] In step S203, the device (200) can generate feature data for life cycle analysis by preprocessing pet registration data and pet management data according to preset preprocessing criteria.

[0115] The preprocessing criteria refer to processing criteria that the device (200) has pre-set to convert pet registration data and pet management data into a format suitable for input to an AI model. The preprocessing criteria may include at least one of a data format unification criterion, a unit conversion criterion, a collection time alignment criterion, a missing data processing criterion, a duplicate data removal criterion, anomaly data processing criterion, and an identification information matching criterion.

[0116] Feature data for life cycle analysis is processed data that an AI model uses as input data when calculating the life cycle status data and care need data of a pet. Feature data for life cycle analysis is not original data that stores pet registration data and pet management data as is, but may be data that the device (200) has organized or converted to suit the purpose of analysis.

[0117] The device (200) can match pet registration data and pet management data based on pet identification information. For example, if the same guardian registers multiple pets through the guardian's terminal (100), the device (200) can identify which pet the health check history, activity record, intake record, or service usage history is related to, and then organize it into the data of the corresponding pet.

[0118] The device (200) can convert data collected in different formats into a unified format. For example, if weight information is entered in kilograms or grams, the device (200) can convert it into a numerical value of the same unit according to a preset unit conversion standard. Additionally, if age information is entered as a date of birth, the device (200) can convert it into an age value or a month value using a reference date.

[0119] The device (200) can sort data collected at different times in chronological order. For example, if health checkup history, biometric data, lifestyle record data, and service usage history data are collected on different dates, the device (200) can sort the data based on the collection date or registration date included in each data to create a data structure according to the flow of time.

[0120] The device (200) can perform processing on missing data, duplicate data, or abnormal data. For example, if life record data for a specific date that needs to be entered through the guardian's terminal (100) is missing, the device (200) can mark the item as a missing value or process it according to a preset correction criterion. In addition, if the same test result is entered duplicately, only one data can be left as an analysis target according to the duplicate data removal criterion.

[0121] The device (200) can generate numeric values, categorical values, time series values, or summary values ​​that can be processed by an AI model using the preprocessing results. For example, the device (200) can convert items such as species, breed, and gender into categorical values, and items such as weight, age, activity level, intake amount, bowel status, amount of sleep, and service usage history into numeric values ​​or summary values.

[0122] The device (200) can generate feature data for life cycle analysis by combining these transformation results. The feature data for life cycle analysis may include at least one of a feature value corresponding to the basic attributes of the pet, a feature value corresponding to the health status, a feature value corresponding to the biological status, a feature value corresponding to the living status, and a feature value corresponding to the service usage status.

[0123] Through this, the device (200) can convert data of various formats collected from the guardian's terminal (100), an external device, or an external service server into consistent analysis data. As a result, analysis errors caused by data format inconsistencies, collection time inconsistencies, duplicate inputs, or missing inputs can be reduced, and the quality of input data for the AI ​​model to calculate the pet's life cycle status and care needs can be improved.

[0124] In step S204, the device (200) can input feature data for life cycle analysis into a pre-trained AI model to produce life cycle status data and care need data of the pet.

[0125] An AI model is an artificial intelligence-based computational model that learns the relationships between feature values ​​contained in input data and generates output data corresponding to the input data. An AI model can be implemented as a neural network model, a decision tree-based model, a random forest model, a gradient boosting model, a regression model, a classification model, or a combination thereof, but is not limited thereto.

[0126] Life cycle status data is data indicating which life cycle status a pet currently belongs to. Life cycle status data can be calculated based on the pet's age, species, breed, weight changes, health history, vital condition, living condition, and service usage status. For example, life cycle status data can be generated as status values ​​corresponding to at least one of the following: growth stage, adulthood, old age, end-of-life care requirement state, or observation requirement state.

[0127] Care need data is data indicating the type or level of care required for a pet. Care need data can be generated as numerical values, grade values, or status values ​​corresponding to at least one of health care need, nutritional care need, activity care need, examination care need, end-of-life care need, and aftercare need.

[0128] The device (200) can provide feature values ​​regarding the basic attributes, health status, biological status, living status, and service usage status of a pet included in feature data for life cycle analysis as input data for an AI model. For example, the device (200) can input at least one of an age value, breed value, weight value, disease history value, health checkup result value, activity level value, intake amount value, bowel status value, sleep amount value, and service usage history value into the AI ​​model.

[0129] The AI ​​model can output life cycle status data and care need data based on input feature data for life cycle analysis. For example, the device (200) can produce life cycle status data corresponding to the elderly stage or the observation need stage for a pet that is old, has a large recent weight change, and has both reduced activity and abnormal health checkup items. Additionally, the device (200) can produce care need data in which the health management need or the health checkup management need is indicated as a relatively high value.

[0130] The device (200) is not limited to a configuration that generates the output result of an AI model as only a single state value. The device (200) may generate at least one of a probability value by life cycle state, a need score by management item, a priority management item, a caution item, or an intermediate score used for service matching.

[0131] For example, the device (200) may calculate a senior age probability value of 0.82 and an adult age probability value of 0.18 as life cycle status data for a specific pet, and calculate a check-up management need score of 85, a nutrition management need score of 70, and an activity management need score of 60 as care need data. These figures and grades are examples for illustrative purposes, and the device (200) may use different score ranges, grade systems, or status classification criteria depending on the embodiment.

[0132] The device (200) is not limited to providing the output results of the AI ​​model directly to the guardian's terminal (100). Life cycle status data and care need data can be used to generate service matching data, generate life cycle measurement management information, generate end-of-life care information, or generate post-care information.

[0133] In one embodiment, the device (200) can generate pet-specific analysis classification data based on life cycle status data and care need data. The analysis classification data is data for classifying the pet's life cycle data analysis content by period, age, species, breed, weight range, or health status.

[0134] Classification by period may include classification by month, season, vaccination cycle, health checkup cycle, service usage cycle, or lifestyle record input cycle. Classification by age may include at least one classification among infancy, growth stage, adulthood, middle age, old age, or end-of-life care requirements. Classification by species may include classification of dog, cat, or other pet species, and classification by breed may include subdivisions within the same species based on breed characteristics.

[0135] When generating analysis classification data, the device (200) can reflect the species, breed, age, weight, and gender included in the pet registration data, as well as the health history data, biometric data, lifestyle record data, and service usage history data included in the pet management data. Accordingly, even when the same life cycle status data is generated, the device (200) can apply different criteria for generating customized information based on the pet's species, breed, age, or time conditions.

[0136] According to this configuration, the device (200) can structure life cycle status data and care need data into analytical classification data that can be used for subsequent service matching and information provision according to criteria by time period, age, type, or breed, rather than using them only as simple status values.

[0137] Through this, the device (200) does not determine the service based solely on fragmentary information input from the guardian's terminal (100), but can calculate the current state of the pet and the level of care required by synthesizing analytical feature values ​​generated from pet registration data and pet management data. As a result, the device (200) can secure objective basic data to perform service matching corresponding to changes in the pet's state.

[0138] In step S205, the device (200) can generate service matching data corresponding to at least one of a well-aging service, an end-of-life care service, a memorial service, and a healing content provision service based on life cycle status data, care needs data, and guardian condition data.

[0139] Service matching data is data generated by the device (200) to determine the service to be provided based on the current status of the pet and the guardian's usage conditions. The service matching data may include at least one of a service type, service identification information, availability of provision, priority of provision, time of provision, notification conditions, and guidance items to be provided to the guardian's terminal (100).

[0140] A well-aging service is a service for managing the health, living conditions, or aging status of a pet. For example, a well-aging service may include at least one of guidance on health checkups, guidance on the use of test kits, guidance on nutritional management, guidance on activity management, guidance on entering daily records, or notifications for regular care.

[0141] End-of-life care services are services provided in response to a pet's end-of-life condition or the level of need for end-of-life care. For example, end-of-life care services may include at least one of guidance on funeral preparations, guidance on transportation support, guidance on information required for guardian input, guidance on the scheduled time of service execution, or end-of-life related notifications.

[0142] A memorial service is a service provided to a guardian in connection with the completion of a pet's funeral or the commencement of a memorial service. For example, the memorial service may include at least one of an electronic memorial space, memorial content, memorial messages, a guestbook, or instructions for using the memorial space.

[0143] The healing content provision service is a content service provided in response to the guardian's memorial activities or post-mortem usage history. For example, the healing content provision service may include at least one of art-based content, meditation-based content, counseling-linked content, or community-linked content. In this case, the healing content provision service is not limited to medical acts of diagnosing or treating the guardian's psychological state, but can be understood as an information provision service to assist the guardian in using the content.

[0144] In one embodiment, the healing content provision service may further include an AI counseling response support service that analyzes the guardian's counseling input to generate counseling type data and counseling response candidate data. The counseling input may include at least one of text, selection items, emotional state selection values, questions related to mourning, records of life changes, reactions to content used, or counseling conversation history entered through the guardian's terminal (100).

[0145] The device (200) can generate counseling type data by analyzing counseling input content according to a pre-learned counseling type analysis model or a pre-set counseling classification criterion. The counseling type data may include at least one of a memorial reminiscence type, a comfort request type, an information guidance type, a content recommendation type, a community linkage type, a counselor linkage need type, and a attention expression detection type.

[0146] The device (200) can generate counseling response candidate data based on counseling type data and guardian condition data. The counseling response candidate data may include at least one of an automatic response sentence candidate, a follow-up question candidate, a healing content candidate, a counseling-linked content candidate, a community guidance candidate, professional counselor linkage guidance information, and caution expression response guidance information.

[0147] In this case, the AI ​​counseling response support service is not limited to medical acts of diagnosing or treating the psychological state of a guardian, but can be understood as an information provision service intended to provide emotional support content, counseling linkage information, or general response information in response to the guardian's counseling input.

[0148] The device (200) can determine the type of service required for a pet using life cycle status data and care need data. For example, if the life cycle status data corresponds to the elderly stage and the care need data indicates a high need for examination management, the device (200) can generate service matching data corresponding to a well-aging service that includes guidance on health checkups or guidance on using test kits.

[0149] The device (200) can select services that are actually available by reflecting guardian condition data together. For example, if the guardian condition data includes budget conditions, available regions, notification reception conditions, and preferred service types, the device (200) can include only services that satisfy those conditions in the service matching data. Additionally, if the guardian does not allow the reception of memorial content, the device (200) can exclude or configure service provision information related to the provision of memorial content in a limited form.

[0150] In one embodiment, the device (200) can generate customized information candidate data based on life cycle status data, care need data, and analysis classification data. The customized information candidate data may refer to candidate data for management guidance, service guidance, or recommendation information that can be provided to the guardian's terminal (100) according to the life cycle analysis content for each pet.

[0151] Personalized information candidate data may include at least one of vaccination scheduled date guidance information, health checkup recommended date guidance information, test kit usage guidance information, nutrition management guidance information, nutritional supplement candidate information, activity management guidance information, hotel usage candidate information, campsite usage candidate information, walking location candidate information, and leisure activity candidate information.

[0152] When generating customized information candidate data, the device (200) may reflect at least one of the following: conditions by time period, conditions by age, conditions by category, conditions by breed, the guardian's available area, the guardian's budget condition, conditions for receiving notifications, and preferred service types. For example, the device (200) may generate vaccination scheduled date guidance information based on vaccination history and age information, and may generate health checkup recommended date guidance information based on health checkup history and the need for checkup management. Additionally, the device (200) may generate nutrition management guidance information or nutritional supplement candidate information based on the need for nutrition management and lifestyle record data, and may generate hotel usage candidate information or campsite usage candidate information based on the guardian's available area and preferred service types.

[0153] In this case, nutritional supplement candidate information or nutritional management guidance information is not limited to prescription information for diagnosing or treating diseases in pets, but can be understood as general management guidance information or service candidate information provided based on life cycle status data, care needs data, and guardian condition data.

[0154] In addition, candidate information for hotel use, candidate information for campsite use, candidate information for walking places, and candidate information for leisure activities can be understood as sub-guide information or candidate information for supplementary services of well-aging services provided in response to the activity management and lifestyle management of pets or the usage conditions of guardians.

[0155] The device (200) may include a candidate satisfying the guardian condition data and the provision availability condition among the customized information candidate data in the service matching data. In this case, the service matching data may include at least one of the service type, service identification information, provision priority, and notification condition, as well as the customized information type, provision target time, provision channel, scheduled time of dispatch, and whether guardian verification is required.

[0156] Additionally, in one embodiment, the device (200) may incorporate reward point data stored corresponding to guardian identification information into the generation of service matching data. The device (200) may check at least one of the quantity of reward point data held, the period of availability, the type of service available, and the usage restriction conditions, and compare this with the cost conditions or availability conditions of the service candidate data.

[0157] If the device (200) satisfies the conditions for the availability of reward point data, it can generate point usage condition data corresponding to at least one of the following: deduction of subscription service costs, deduction of well-aging service usage costs, deduction of health check-up related service usage costs, deduction of end-of-life care service usage costs, deduction of memorial service usage costs, or deduction of pet supplies purchase costs.

[0158] At this time, the service matching data may further include at least one of service identification information, provision priority, and notification conditions, as well as the applicability of reward point data, the quantity of applicable points, the estimated deduction amount, or the remaining payment amount.

[0159] In this case, information regarding subscription service costs, pet supply purchase costs, or other service usage costs may be understood as cost conditions, payment conditions, or additional usage conditions included in the service matching data, and is not limited to essential service types separate from well-aging services, end-of-life care services, memorial services, or healing content provision services.

[0160] Additionally, the device (200) may incorporate prepaid deposit asset management data stored in correspondence with guardian identification information into the generation of service matching data. In this case, the prepaid deposit asset management data may be used as part of guardian condition data representing the guardian's service usage conditions, budget conditions, or cost burden conditions.

[0161] The device (200) can check at least one of the deposit balance information, available service type, available period, and usage restriction conditions included in the prepaid deposit asset management data, and compare it with the cost conditions or availability conditions of the service candidate data.

[0162] If the prepaid deposit asset management data satisfies the conditions for availability, the device (200) can generate deposit asset usage condition data corresponding to at least one of a deduction of subscription service costs, a deduction of well-aging service usage costs, a deduction of health check-up related service usage costs, a deduction of end-of-life care service usage costs, a deduction of funeral service usage costs, a deduction of memorial service usage costs, a deduction of healing content provision service usage costs, or a deduction of pet supplies purchase costs.

[0163] In this case, the service matching data may further include at least one of the service identification information, provision priority, and notification conditions, as well as the applicability of prepaid deposit asset management data, the applicable deposit range, the estimated deduction amount, or the remaining deposit information.

[0164] The device (200) is not limited to a configuration that selects only one service. Multiple services may be matched together depending on the life cycle status and care needs of the pet. For example, the device (200) may generate service matching data that includes guidance on regular checkups, guidance on nutritional management, and notifications for entering daily life records for elderly pets. Additionally, if the level of need for end-of-life care is also identified, guidance on end-of-life care services may be additionally included.

[0165] When generating service matching data, the device (200) may include service-specific provision times or notification conditions. For example, the device (200) may configure the service matching data so that health checkup guidance is provided at the time of the next recommended checkup, life record input guidance is provided at regular intervals, and end-of-life care service guidance is provided at the time when the guardian input required information is confirmed.

[0166] Through this, the device (200) can determine the target for service provision by considering the condition of the pet and the guardian's usage conditions together. As a result, the service provision information provided to the guardian's terminal (100) is not limited to a simple product list or fixed guidance information, but can be composed of customized information corresponding to the pet's life cycle status and care needs.

[0167] For a detailed explanation regarding this, refer to Fig. 7.

[0168] In step S206, the device (200) can provide service provision information corresponding to service matching data to the guardian's terminal (100).

[0169] Service provision information is guidance information provided to a guardian according to service matching data generated by the device (200). The service provision information may include at least one of a service type, service name, purpose of provision, time of provision, availability conditions, expected usage procedure, notification information, and items required for guardian input.

[0170] The device (200) may provide service provision information through at least one of an application, web page, notification message, text message, email, external messaging server, external messenger server, or SNS server installed on the guardian's terminal (100). In this case, the guardian's terminal (100) may display the service provision information received directly from the device (200) or received through an external messaging server, external messenger server, or SNS server on a screen or output it in the form of a notification.

[0171] For example, the device (200) can provide service provision information to a guardian's terminal (100), including at least one of guidance on scheduled vaccination dates, guidance on recommended health check-up dates, guidance on test kit usage, guidance on nutritional management, guidance on nutritional supplement candidates, guidance on hotel usage candidates, guidance on camping site usage candidates, or guidance on leisure activity candidates, according to life cycle status data, care need data, and analysis classification data.

[0172] Additionally, the device (200) may provide service provision information at a specific date, a specific period, the time when the scheduled vaccination date arrives, the time when the recommended health checkup date arrives, the time when lifestyle records are missing, the time when the nutritional management need criteria are met, or the time when the leisure activity recommendation conditions are met, depending on the notification reception conditions included in the guardian condition data.

[0173] In one embodiment, the device (200) checks the notification reception consent information, external message reception consent information, reception channel information, and reception time conditions included in the guardian condition data, and if the preset automatic sending criteria are satisfied, it can convert the service provision information into automatic sending information. The automatic sending information may include at least one of recipient identification information, sending channel information, scheduled sending time, message subject, message body, connection screen identification information, and service detail screen link information.

[0174] The device (200) can provide automatic sending information to an application installed on the guardian's terminal (100), or transmit it to an external messaging server, an external messenger server, or an SNS server so that it is displayed on the guardian's terminal (100).

[0175] For example, the device (200) can generate a vaccination guidance message as automatic sending information when the scheduled vaccination date arrives, and can generate a health checkup guidance message as automatic sending information when the recommended health checkup date arrives. Additionally, the device (200) can generate a nutrition management guidance or nutritional supplement candidate guidance message when the need for nutrition management is above a preset standard, and can generate a hotel, campsite, or leisure activity candidate guidance message when the guardian's available area and preferred service type are identified.

[0176] The device (200) can store the sending results of the automatic sending information by correlating them with guardian identification information, pet identification information, sending channel, sending time, and sending status information. The sending status information may include at least one of scheduled sending, sent completed, sent failed, receipt confirmation, viewing confirmation, or guardian response confirmation.

[0177] According to this configuration, the device (200) can classify the life cycle analysis content for each pet according to criteria by time period, age, species, or breed, and automatically provide customized information related to vaccination, health check-ups, nutritional management, nutritional supplement candidates, hotels, campsites, or leisure activities corresponding to the classification to the guardian's terminal (100). Accordingly, the guardian can check information corresponding to the pet's condition and time period through an application or external messaging channel without having to search for services separately.

[0178] The device (200) is not limited to a configuration that provides service provision information as a simple fixed list. The device (200) can adjust the order, display status, or notification method of service provision information displayed on the guardian's terminal (100) according to the provision priority, availability, timing of provision, and notification conditions included in the service matching data.

[0179] For example, the device (200) checks the budget conditions and available regions in the guardian condition data, and then prioritizes displaying available well-aging service information. Services for which the available regions are not met may not be displayed or may be provided as separate guidance information. Additionally, the device (200) may provide service provision information related to memorial services or healing content provision services to the guardian's terminal (100) only if the guardian has allowed the reception of memorial content.

[0180] After receiving service provision information, the guardian's terminal (100) may display at least one of a service inquiry screen, a service details screen, a reservation screen, a notification screen, a usage result input screen, a counseling input screen, a counseling response screen, or a healing content recommendation screen. The guardian can check the service provision information through the guardian's terminal (100) and perform service selection, reservation, rejection, suspension, input of additional information, input of counseling content, or selection of healing content.

[0181] In one embodiment, the device (200) can receive counseling input content from a guardian through a counseling input screen displayed on the guardian's terminal (100). The counseling input content may include at least one of a counseling sentence entered by the guardian, selectable emotional state information, information on the pet being mourned, counseling purpose information, preferred response method information, and previous counseling conversation history.

[0182] The device (200) can generate counseling type data by inputting counseling input content into a counseling type analysis model or comparing it with preset counseling classification criteria. The counseling type analysis model may be an artificial intelligence model trained to produce counseling type data using at least one of sentence expressions, emotional expressions, repetitive expressions, question types, keywords related to mourning, and the guardian's response history of the counseling input content.

[0183] The device (200) can select counseling response candidate data corresponding to counseling type data and generate counseling response information based on the selected counseling response candidate data. The counseling response information may include at least one of an automatic response sentence, a follow-up question, healing content recommendation information, meditation content recommendation information, art-based content recommendation information, community guidance information, professional counselor referral guidance information, or attention expression response guidance information.

[0184] The device (200) can provide the generated counseling response information to a counseling response screen through an application, web browser, or platform access screen installed on the guardian's terminal (100). In this case, the guardian's terminal (100) can display the counseling response information in at least one form among an interactive message, card-type guidance information, a content recommendation list, or a counselor linkage guidance button.

[0185] If the device (200) determines that expressions of self-harm, abuse, violence, extreme pain, or expressions requiring immediate action in the counseling input content satisfy preset warning expression criteria, it may generate counseling response information including guidance on connecting to a professional counselor, guidance on a guardian support agency, or guidance on requesting emergency help instead of a general automatic response sentence.

[0186] According to this configuration, the device (200) can analyze the counseling content entered by the guardian through the guardian's terminal (100) by type and automatically provide response information or healing content information corresponding to the counseling type. Accordingly, the healing content provision service is not limited to simply providing fixed content, but can be provided as a customized response support service corresponding to the guardian's counseling input content and usage conditions.

[0187] Additionally, in one embodiment, the guardian's terminal (100) may display at least one of point balance information corresponding to the reward point data, point accumulation history information, information on services where points can be used, information on the amount to which points can be applied, and a point usage selection item, along with service provision information. The guardian may choose to apply the reward point data to at least one of the subscription service cost, service usage cost, or pet supply purchase cost through the guardian's terminal (100).

[0188] When the device (200) receives point usage selection information from the guardian's terminal (100), it checks the point usage condition data and can deduct the amount of available points corresponding to the service usage cost or the cost of purchasing goods. The device (200) stores the point usage result corresponding to the guardian identification information and can provide point usage completion information or remaining point information to the guardian's terminal (100).

[0189] In one embodiment, the device (200) may generate or update prepaid deposit asset management data corresponding to guardian identification information. The prepaid deposit asset management data may include at least one of deposit payment history, deposit usage history, balance information, refund history, service-specific deduction history, and deposit status information related to subscription service costs, well-aging service costs, end-of-life care service costs, funeral service costs, memorial service costs, healing content provision service costs, other service usage costs, or pet supply purchase costs.

[0190] The device (200) may record at least one of deposit history identification information, transaction identification information, service deduction history information, balance change history information, or integrity verification value corresponding to prepaid deposit asset management data on a blockchain network or a distributed ledger network. In this case, the device (200) is not limited to directly storing the original text of the deposit amount, balance information, or guardian identification information on the blockchain network, but may record a hash value or verification value corresponding to the original text data.

[0191] The device (200) can provide a deposit asset information provision screen including at least one of deposit payment history, deposit usage history, balance information, service-specific deduction history, refund history, point usage history, and integrity verification results through an application, web browser, or platform access screen installed on the guardian's terminal (100).

[0192] The guardian's terminal (100) can receive a request to view prepaid deposit asset management data, a request to check usage history by period, a request to check deduction details by service, or a request to check integrity verification results through a deposit asset information provision screen. When the device (200) receives a request to view from the guardian's terminal (100), it can view the prepaid deposit asset management data corresponding to the guardian identification information and provide the verification result generated by referencing a verification value recorded in a blockchain network or a distributed ledger network to the guardian's terminal (100).

[0193] According to this configuration, the device (200) can provide the guardian with the ability to check the history of payment, use, deduction, and balance change of assets deposited in advance through the guardian's terminal (100). Additionally, the device (200) can enhance the reliability of prepaid asset management by allowing verification of whether the deposit history has been tampered with using a hash value or a verification value, without disclosing the original text of the prepaid asset management data to the outside.

[0194] Through this, the device (200) can specifically provide service information corresponding to the pet's life cycle status and care needs through the guardian's terminal (100). Accordingly, the guardian can check service information that matches the pet's current status and the guardian's conditions, and the device (200) can subsequently collect response data such as the guardian's inquiry, selection, reservation, or refusal.

[0195] In step S207, the device (200) can generate feedback data for training an AI model based on response data to service provision information received from the guardian's terminal (100) and usage result data corresponding to the service provision information.

[0196] Response data is data representing the result of an action or selection performed by a guardian regarding service provision information through the guardian's terminal (100). For example, response data may include at least one of whether service provision information is viewed, the time of viewing, whether selection is made, whether reservation is made, whether rejection is made, whether suspension is made, whether notification is checked, whether additional information is entered, and satisfaction input information. In one embodiment, response data may further include at least one of whether automatic sending information is viewed, whether a link is selected, whether access to a service detail screen is made, whether rejection is made, or whether notification is blocked.

[0197] Additionally, in one embodiment, the response data may further include at least one of whether the counseling response information was viewed, whether additional questions were entered, whether healing content was selected, whether counselor linkage was selected, whether satisfaction was entered, whether the response was discontinued, or whether notifications were blocked. The usage result data may include at least one of the content usage time after the provision of counseling response information, the duration of the counseling session, information on changes in counseling type, subsequent input by the guardian, or whether the counseling linkage service was used.

[0198] Usage result data is result data generated after the service corresponding to the service provision information has been actually used. For example, the usage result data may include at least one of the following: whether the service was completed, the time of service use, the location of service use, the number of times the service was used, test results, health checkup results, information on changes in lifestyle records, information on service reservation completion, information on end-of-life care service processing, information on pet funeral completion, and information on the commencement of memorial service.

[0199] Feedback data for AI model training is data that organizes response and usage result data so that it can be used for the subsequent training or updating of the AI ​​model. Feedback data for AI model training can represent the correspondence between service provision information, guardian responses, and actual usage results. For example, feedback data for AI model training may include what service provision information was provided regarding certain life cycle status data and care needs data, whether the guardian selected the corresponding service, and what the actual usage results were.

[0200] The device (200) can receive response data from the guardian's terminal (100). For example, if the guardian views and reserves a health checkup guide on the guardian's terminal (100), the device (200) can receive the identification information of the service provision information, the time of viewing, whether a reservation was made, and the time of reservation as response data. Additionally, even if the guardian rejects the service provision information or closes the notification, the device (200) can receive the rejection status or the result of non-selection as response data.

[0201] The device (200) can collect usage result data corresponding to service provision information. For example, if a health checkup is completed in accordance with the health checkup guide, the device (200) can collect information on whether the checkup was completed, the date of the checkup, and the checkup result as usage result data. If a test kit is used in accordance with the test kit usage guide, the device (200) can collect information on whether the test was completed and the test results as usage result data. If an end-of-life care service or a memorial service is used, the device (200) can collect information on whether the service processing was completed, information on the completion of the burial, or information on the start of the memorial service as usage result data.

[0202] The device (200) can correspond reaction data and usage result data to each other based on identification information of service provision information, pet identification information, or guardian identification information. Through this, the device (200) can organize into a single data structure whether the service information provided according to the condition of a specific pet was actually selected or used, and what result occurred after use.

[0203] The device (200) can extract or convert items from response data and usage result data that can be used for training an AI model. For example, the device (200) can convert whether a service is selected into a binary value, whether the service is completed into a status value, and the time interval between the time of service use and the time of the usage result occurrence into a numeric value. In addition, the device (200) can convert satisfaction input information or the usage result status into a preset label value.

[0204] The device (200) can generate feedback data for training an AI model by matching the converted data with the life cycle status data, care need data, and service matching data at the time of service provision. For example, the device (200) can generate a single training feedback data result in which a health checkup guide is provided to a pet for which life cycle status data corresponding to old age and care need data with a high need for checkup management have been calculated, and the guardian selects the service and completes the checkup.

[0205] The device (200) is not limited to discarding the data as is even if some response data or usage result data is missing. The device (200) may select only items available for learning based on preset data validity criteria, or display incomplete data as a separate status value. For example, if service provision information has been retrieved but usage result data has not yet been collected, the device (200) may manage the data as feedback data in an observation state.

[0206] Through this, the device (200) can organize the guardian's reaction after service provision and the actual usage results into a form that can be used for AI model training. Accordingly, the device (200) does not stop at simply providing services, but can reflect in subsequent learning which service was selected or used for a pet in a certain state, and can establish a data base to continuously improve the accuracy of service matching.

[0207] For a detailed explanation regarding this, refer to Fig. 3.

[0208] In step S208, the device (200) can store feedback data for AI model training in a pet life cycle learning database.

[0209] The pet lifecycle learning database is a data storage that stores data for the device (200) to use for training, verifying, or updating an AI model. The pet lifecycle learning database may be implemented in the internal storage of the device (200), or may be implemented as an external database server, cloud storage, or distributed storage structure capable of communicating with the device (200).

[0210] The device (200) can store feedback data for AI model training in a pet lifecycle learning database by correlating it with at least one of pet identification information, guardian identification information, service identification information, lifecycle status data, care need data, and creation time information. Through this, the device (200) can use the relationship between the status of a specific pet, the provided service information, the guardian's reaction, and the actual usage result for subsequent analysis.

[0211] The device (200) can store feedback data for AI model training by separating it into original data and processed data. For example, the device (200) can store response data and usage result data received from the guardian's terminal (100) as original data, and data converted into numerical values, category values, status values, or label values ​​suitable for AI model training as processed data. By separating the storage structure in this way, the device (200) can use processed data during the AI ​​model training process, while also being able to refer to the original data if necessary to verify the basis for data generation.

[0212] When storing feedback data for training an AI model, the device (200) may store at least one of the data creation time, collection source, data validity status, and update history together. For example, the device (200) may store the time when service provision information is provided, the time when a guardian selects the information, the time when the service is actually used, and the time when the usage result is collected, respectively. Through this, the device (200) can analyze the service provision results over time.

[0213] The device (200) can check whether the data to be stored is duplicated or whether identification information is missing before the feedback data for AI model training is stored in the pet life cycle learning database. For example, if the same service usage result is transmitted multiple times, the device (200) can distinguish duplicate data by comparing the service identification information, pet identification information, and the time of completion of use. Additionally, data with missing pet identification information can be stored as a separate verification target or excluded from the training target.

[0214] The device (200) can use feedback data for training AI models stored in a pet life cycle learning database for subsequent AI model training or updating. For example, the device (200) can reflect as training data whether a service provided in response to a specific life cycle state and care needs was selected by the guardian, actually used, and whether the results after use were good.

[0215] The device (200) can use data stored in a pet life cycle learning database for training an AI model according to a set period or preset update conditions. For example, the device (200) can update the AI ​​model using the stored feedback data when a certain amount of feedback data for training a new AI model has accumulated, when results of service usage for a specific period have been collected, or when a request for model update is entered by an administrator.

[0216] The device (200) may perform processing for the protection of personal or sensitive information when storing feedback data for AI model training. For example, the device (200) may store guardian identification information and pet management data separately, or convert guardian identification information into an alternative identifier and store it. This processing is intended to reduce unnecessary exposure of identification information during the process in which information collected through the guardian's terminal (100) is used for AI model training.

[0217] In one embodiment, the device (200) may perform encryption processing on at least one of guardian identification information, pet identification information, health history data, biometric data, life record data, service usage history data, response data, usage result data, and feedback data for AI model training among the data stored in the pet life cycle learning database. The encryption processing may include at least one of symmetric key encryption, public key encryption, key separation storage, access control, and de-identification processing of identification information.

[0218] At this time, the feedback data for training the AI ​​model may include counseling type data, counseling response information, guardian's reaction data to the counseling response information, and counseling usage result data. The device (200) may de-identify or encrypt information that can be directly linked to guardian identification information among counseling input content or counseling conversation history, and then use it to update the counseling type analysis model or counseling response candidate selection criteria.

[0219] When storing counseling-related feedback data, the device (200) can store counseling input original text, counseling type data, counseling response information, and guardian reaction data separately. Additionally, the device (200) can control whether to store or use counseling-related feedback data for learning based on the scope of the guardian's consent, the counseling data retention period, whether learning use is permitted, and whether consent is withdrawn.

[0220] The device (200) is not limited to directly storing the original data to be stored or the original personal information text on a blockchain network. The device (200) may generate at least one of a hash value, storage time information, data source information, access history information, change history information, or integrity verification value corresponding to the data to be stored, and record the generated value on a blockchain network or a distributed ledger network.

[0221] When a request to query, modify, or delete stored data occurs, the device (200) can generate a data integrity verification result by comparing the current hash value corresponding to the data stored in the database with a verification value recorded in a blockchain network or a distributed ledger network. In this case, the deletion request may be a request regarding the deletion, de-identification, or access restriction processing of original data stored in the database, and the hash value or integrity verification value recorded in the blockchain network or the distributed ledger network may be managed as verification data that cannot restore the original data.

[0222] At this time, the data integrity verification result may include at least one status value among normal, suspected change, unverifiable, or need for verification.

[0223] Through this, the device (200) manages customer-related data, such as pet health history data, lifestyle record data, service usage history data, and feedback data for AI model training, by encrypting it, and can verify whether the storage or change history has been tampered with through a blockchain-based verification value. Accordingly, the device (200) can enhance the security and integrity of the data stored in the pet life cycle learning database.

[0224] Consequently, the device (200) can accumulate response data and usage result data generated after service provision as data that can be used for subsequent training of an AI model, rather than managing them only as one-time records. Accordingly, the device (200) can continuously reflect the results of service provision based on the pet's life cycle status and care needs, and subsequently improve the accuracy of service matching for pets in the same or similar state.

[0226] Meanwhile, the AI ​​model can be trained based on a pet life cycle learning database and configured to generate life cycle status data and care need data as output data using feature data for life cycle analysis as input data.

[0227] Here, the learning of the AI ​​model may refer to the process of setting or updating the parameters, weights, classification criteria, or prediction criteria of the model so that the device (200) can calculate the correspondence between the input data and the output data using the learning data stored in the pet life cycle learning database.

[0228] The pet life cycle learning database may be a database that stores at least one of the following in a form usable for learning: pet-specific registration information, health history, biological condition, life record, service usage history, service provision results, and feedback data. The device (200) can learn which life cycle status data and care need data correspond to a specific combination of input characteristics by using the data stored in the pet life cycle learning database.

[0229] For example, the pet life cycle learning database may store life cycle status values ​​or care need values ​​corresponding to the pet's species, breed, age, weight, health check results, disease history, activity level, intake, defecation status, sleep amount, and service usage history. The device (200) can use this data to train an AI model to classify the pet's current status or calculate the level of care needed.

[0230] Feature data for life cycle analysis is data input into an AI model. Feature data for life cycle analysis may include numerical values, categorical values, time series values, or summary values ​​generated by preprocessing pet registration data and pet management data. For example, feature data for life cycle analysis may include at least one of age values, breed values, weight values, disease history values, health checkup results, activity levels, intake values, bowel status values, sleep levels, and service usage history values.

[0231] Life cycle status data is one of the data outputs by an AI model and can represent the current life cycle status of a pet. Life cycle status data may be generated as a single status value, or as probability values, scores, or grade values ​​for multiple status candidates.

[0232] Care needs data is one of the data outputs by an AI model and can represent the care items or levels required for a pet. Care needs data may include need values, grade values, or status values ​​related to health care, nutrition care, activity care, check-up care, end-of-life care, or aftercare.

[0233] The device (200) can input feature data for life cycle analysis into an AI model and receive life cycle status data and care need data from the AI ​​model. For example, the device (200) can input feature data for life cycle analysis, including age, weight change, disease history, health checkup results, and lifestyle records regarding a specific pet, into the AI ​​model and produce life cycle status data corresponding to the elderly state and care need data with a high need for checkup management from the AI ​​model.

[0234] The AI ​​model is not limited to a specific algorithm. The AI ​​model can be implemented as a neural network model, a decision tree-based model, a random forest model, a gradient boosting model, a regression model, a classification model, or a combination thereof. The device (200) can select a suitable model structure according to the structure of the data stored in the pet life cycle learning database, the learning purpose, and the format of the output data.

[0235] Additionally, the device (200) can retrain or update the AI ​​model according to preset update conditions when feedback data for AI model training generated after service provision is stored in the pet life cycle learning database. For example, the device (200) can perform training of the AI ​​model when a certain number of feedback data is accumulated, at regular intervals, or when a request for model update is entered by an administrator.

[0236] Through this, the device (200) can calculate life cycle status data and care need data by utilizing data accumulated in the pet life cycle learning database and the learning results of the AI ​​model, rather than determining the service based solely on fragmentary input information collected from the guardian's terminal (100). Accordingly, as the pet's condition changes and service usage results accumulate, the device (200) can increase the reliability of the analysis results used for subsequent service matching.

[0237] As a result, the device (200) can analyze the life cycle status and care needs based on the pet's registration information, management history, and guardian conditions, and continuously perform corresponding service provision and feedback learning.

[0238] FIG. 3 is a flowchart illustrating the process of generating feedback data for training an AI model according to one embodiment.

[0239] Referring to FIG. 3, first, in step S301, the device (200) can generate service usage characteristic data by extracting service identification information, whether the service is selected, the time of service use, and information on changes in the pet's condition after service use from response data and usage result data.

[0240] Service identification information is information for distinguishing which service among the service provision information provided by the device (200) to the guardian's terminal (100) has been presented to or actually used by the guardian. The service identification information may include at least one of a service code, a service name, a service type, service provider information, a service provision cycle, a service provision information dispatch identifier, or a service reservation number.

[0241] The service selection status is data indicating whether the guardian has selected the service after checking the service provision information provided through the guardian's terminal (100). The service selection status may be recorded as at least one status value among selection, non-selection, rejection, suspension, reservation, cancellation, or request for additional information.

[0242] The service usage time is time information indicating the point in time when the service corresponding to the service provision information is actually used, reserved, or completed. The service usage time may be set to at least one of the service reservation time, service start time, service completion time, test kit usage time, health checkup completion time, end-of-life care service application time, or memorial service start time.

[0243] Information on changes in the pet's condition after using the service is information indicating whether the pet's health condition, test results, or lifestyle patterns have changed after using the service. Information on changes in the pet's condition after using the service may include at least one of changes in health checkup results, changes in test values, changes in weight, changes in activity level, changes in intake amount, changes in bowel condition, changes in sleep amount, changes in disease condition, or observation records entered through the guardian's terminal (100).

[0244] The device (200) can extract service identification information and service selection status from response data received from the guardian's terminal (100). For example, if the guardian views a health checkup guide displayed on the guardian's terminal (100) and then selects a reservation button, the device (200) can extract service identification information corresponding to the health checkup guide and the reservation status as service selection status. Even if the guardian does not check the notification or selects to decline, the device (200) can extract the unconfirmed or declined status as service selection status.

[0245] The device (200) can extract the time of service use from the usage result data. For example, if a test kit service is provided, the device (200) can extract the time of completion of test kit use or the time of registration of test results as the time of service use. If a health checkup service is provided, the device (200) can extract the date of completion of the checkup or the date of receipt of the checkup results as the time of service use.

[0246] The device (200) can extract information on changes in the pet's condition after using the service from usage result data or pet management data linked to the usage result data. For example, if weight, intake amount, or bowel status is newly entered after nutritional management guidance is provided, the device (200) can extract the corresponding input value as information on changes in the pet's condition after using the service. Additionally, if test results are registered after using the health checkup service, the device (200) can extract the test values ​​or the status of the test results as information on changes in the pet's condition after using the service.

[0247] The device (200) can generate service usage characteristic data by correlating extracted service identification information, service selection status, service usage time, and pet condition change information after service usage. The service usage characteristic data is data indicating which pet a specific service was provided to, how the guardian reacted to the service, and what condition change was confirmed after actual use.

[0248] For example, the device (200) can generate these items as a single service usage characteristic data when the service identification information is a health checkup service, the service selection status is a reservation completion, the service usage time is July 10, 2026, and the pet condition change information after service usage is the registration of test results and observation of reduced activity.

[0249] When generating service usage characteristic data, the device (200) can correspond service identification information, pet identification information, and guardian identification information together. Through this, the device (200) can distinguish which pet the service usage result is for, even if the same guardian has registered multiple pets.

[0250] The device (200) may configure service usage characteristic data into at least one of a numeric value, a category value, a time value, or a status value. For example, whether a service is selected may be converted into a binary value or a status value indicating whether a selection was made, the time of service use may be converted into a date value or a value of days elapsed relative to a reference date, and information on changes in the pet's condition after service use may be converted into whether a change occurred or a status value by change item.

[0251] Through this, the device (200) can not only store the reactions generated through the guardian's terminal (100) and the actual service usage results as a simple history, but also organize them into service usage characteristic data that can be used for AI model training. Accordingly, the device (200) can subsequently analyze whether the provided service was selected, actually used, or if there was a change in state after use in a specific life cycle state or care need.

[0252] In step S302, the device (200) can match information on changes in the pet's condition after using the service with pet registration data, health history data, biometric data and life record data and a pre-assigned pet identifier.

[0253] A pre-assigned pet identifier is unique identification information assigned or managed by the device (200) to distinguish each pet within the platform. The pet identifier can be generated when a pet is registered through the guardian's terminal (100) and can correspond in common to pet registration data, health history data, biometric data, lifestyle record data, and service usage results.

[0254] The device (200) can compare the pet identifier included in the pet status change information after using the service with the pet identifier stored in the pet registration data. If a matching pet identifier is identified, the device (200) can correlate the status change information with the health history data, biometric data, and life record data of the same pet.

[0255] For example, if the first pet and the second pet are registered together on the guardian's terminal (100), the device (200) can check which pet is related to the health checkup results, test kit results, weight change records, or activity change records based on the pet identifier. When the test results are received after using the health checkup service for the first pet, the device (200) can correlate the test results with the first pet's pet registration data, health history data, biometric data, and lifestyle record data.

[0256] The device (200) can check for correspondence using guardian identification information, service identification information, service usage time, reservation information, or test result registration information even for data that does not directly include a pet identifier. For example, if a guardian reserves a test kit service for a specific pet on the guardian's terminal (100), the device (200) can compare the reservation information and the test result upon receiving the test result to correspond to the data of the pet.

[0257] If identification information is missing or conflicts occur during the response process, the device (200) may classify the data as a verification target rather than using it as a learning target immediately. For example, if the same test result is linked to multiple pet identifiers simultaneously, the device (200) may provide verification request information to the guardian's terminal (100) or select the most reliable response relationship according to pre-set response criteria.

[0258] Through this, the device (200) can clearly distinguish which pet the status change information confirmed after using the service is the change of. Accordingly, even when the same guardian manages multiple pets, data confusion can be reduced and the reliability of feedback data for AI model training can be increased.

[0259] In step S303, the device (200) can generate life cycle history data for each pet by arranging the response results in the order of collection time.

[0260] The response result refers to the result of connecting information on changes in the pet's condition after using the service with pet registration data, health history data, biometric data, and lifestyle record data based on the pet identifier. The device (200) can sort the data in chronological order based on the time of collection, time of registration, time of measurement, or time of service use of each data included in the response result.

[0261] The pet-specific life cycle history data is historical data that includes registration information, health status, biological status, living status, and information on changes after using the service, accumulated over time for a specific pet. The pet-specific life cycle history data can be used by the device (200) to compare changes in status before and after using the service, or to generate basic data for feedback data for training an AI model.

[0262] For example, the device (200) can arrange the weight record of June 1, 2026, the activity record of June 5, 2026, the health checkup result of June 10, 2026, the service usage time of June 15, 2026, and the test result of June 30, 2026 for the first pet in chronological order. The device (200) can generate the data arranged in this way as life cycle history data of the first pet.

[0263] If the time of data collection is unclear, the device (200) may use the time of registration, the time of service completion, or the time of creation provided by an external system as a secondary standard. For example, if a record of defecation status is entered through the guardian's terminal (100) but the actual time of observation is not entered, the device (200) may set the time of entry of the record as the time of collection.

[0264] The device (200) can store data by classifying it according to data type when multiple data are collected for the same pet at the same time or at close times. For example, if weight records, intake records, and health check results are collected together on the same date, the device (200) can organize each data into different items belonging to the same collection period.

[0265] Through this, the device (200) can organize the status information of the pet into historical data that reflects the chronological order, rather than managing it as fragmentary individual records. Accordingly, the device (200) can compare the previous state and the subsequent state based on the time of service use, and can more clearly analyze the direction and trend of status changes for each pet.

[0266] In step S304, the device (200) can exclude missing data, abnormal data, and data that does not satisfy the observation period criteria from the lifecycle history data according to preset exclusion criteria.

[0267] The exclusion criteria are criteria for the device (200) to select data unsuitable for training or analysis in order to ensure the quality of feedback data for training an AI model. The exclusion criteria may include at least one of a missing data criterion, an abnormal data criterion, and an observation period criterion.

[0268] Missing data refers to data in which some of the items that should be included in the life cycle history data are missing. For example, if the time of service use is confirmed but there are no examination results, weight records, or lifestyle records after service use, the device (200) may determine that the data is missing data. Additionally, data may be classified as missing data if there is a lack of data in either the reference interval or the observation interval required for comparison before and after service use.

[0269] Abnormal data refers to data that deviates from a preset normal range or data format standard. For example, if the pet's weight value deviates from the inputtable range, the activity value exceeds the allowable range of the measuring device, or if contradictory test results are duplicated at the same time, the device (200) may determine that the data is abnormal data.

[0270] Data that does not satisfy the observation period criteria refers to data for which sufficient data for a period of time has not been secured to determine changes in state before and after the use of the service. For example, if health history data, biometric data, or lifestyle record data is not collected during a preset observation period after the use of the service, the device (200) may classify such data as data that does not satisfy the observation period criteria.

[0271] The device (200) may exclude missing data, abnormal data, and data that does not satisfy the observation period criteria from the lifecycle history data, or manage them separately by assigning a status value that excludes them from the learning target. In this case, exclusion is not limited to deleting the original data, but may include processing to separate it from the analysis target so that it is not used to generate feedback data for AI model training.

[0272] For example, the device (200) may exclude from the learning target data for which test results were not collected after using the health checkup service, data for which it is difficult to determine changes in condition because the observation period after using the service is only one day, or data for which weight values ​​were entered abnormally. On the other hand, even if some items are missing, if the pre-set minimum item criteria are satisfied, the data may be maintained as a target for correction or limited analysis.

[0273] The device (200) may store the reason for exclusion along with the excluded data. For example, the device (200) may record which reason a specific data was excluded due to, among missing values, outliers, insufficient observation period, or discrepancy in identification information. Through this, the device (200) may subsequently utilize the reason for exclusion for data quality verification, re-collection requests, or additional input requests via the guardian's terminal (100).

[0274] Through this, the device (200) can avoid directly reflecting incomplete or unreliable data in the generation of feedback data for AI model training. Accordingly, errors in judging state changes before and after service use can be reduced, and the consistency and reliability of the data used for AI model training can be increased.

[0275] In step S305, the device (200) can divide the life cycle history data into reference interval data prior to the service usage time and observation interval data after the service usage time.

[0276] Reference interval data refers to life cycle history data collected prior to the point of service use, used to indicate the condition of the pet prior to service use. Reference interval data may include health history data, biometric data, and lifestyle record data collected during a period preceding the point of service use.

[0277] Observation interval data is data used to check changes in the condition of a pet after using the service, among the life cycle history data collected after the time of service use. The observation interval data may include at least one of the following: health checkup results, test results, weight records, activity records, intake records, defecation records, sleep records, and observation records entered through the guardian's terminal (100) after using the service.

[0278] The device (200) can set the time of service use as the reference date and divide the life cycle history data into reference interval data and observation interval data according to a preset period range. For example, the device (200) can set the 30 days prior to the service use date as the reference interval and the 30 days after the service use date as the observation interval. The period range may be set differently depending on the service type, the life cycle status of the pet, the need for care, or the data collection cycle.

[0279] For example, in the case of a health checkup service, the device (200) may classify weight, activity level, intake, and lifestyle records prior to the health checkup date as reference interval data, and test results, weight changes, and lifestyle records after the health checkup date as observation interval data. In the case of a test kit service, the device (200) may classify health status records prior to the use of the test kit as reference interval data, and lifestyle records or additional test results after the registration of test results as observation interval data.

[0280] When distinguishing between reference interval data and observation interval data, the device (200) can determine how far apart the collection time of each data is from the service usage time. For example, data collected too long ago before the service usage time may be excluded from the reference interval data because it is difficult to represent the current state, and data collected for too short a period immediately after service usage may be excluded from the observation interval data or managed in a pending state if it is not sufficient for judging a change in state.

[0281] The device (200) is not limited to generating one reference interval data and one observation interval data for a single service usage time. If the service is used multiple times, the device (200) can distinguish the reference interval data and the observation interval data based on each service usage time. Through this, multiple service usage cases for the same pet can be managed as individual learning cases.

[0282] Through this, the device (200) can clearly distinguish between the state before service use and the state after service use based on time. Accordingly, the device (200) can provide a structure in which changes after service use can be compared with the state before service use, and in subsequent processing, the amount of change in health status, the amount of change in test results, and the amount of change in lifestyle patterns can be calculated based on more consistent standards.

[0283] In step S306, the device (200) can calculate the amount of change in health status, the amount of change in test results, and the amount of change in lifestyle patterns by comparing reference interval data and observation interval data.

[0284] The amount of change in health status is data representing the difference in health status before and after the time of service use. The amount of change in health status can be calculated using the degree of change in at least one of the following: disease status, occurrence of symptoms, medication status, medical history, guardian observation records, or health status score.

[0285] The amount of change in test results is data representing the difference in test results before and after the point of service use. The amount of change in test results can be calculated using the degree of change of at least one of the following: health checkup results, test kit results, weight measurements, body temperature measurements, blood test results, urine test results, or other measurements.

[0286] The amount of change in lifestyle patterns is data indicating the difference in daily management status or lifestyle habits before and after the time of service use. The amount of change in lifestyle patterns can be calculated using the degree of change in at least one of the following: activity level, intake, bowel status, urination status, sleep amount, walking time, walking distance, feeding cycle, or the guardian's lifestyle record input value.

[0287] The device (200) can compare the representative value for each item included in the reference interval data with the representative value for each item included in the observation interval data. The representative value can be set to at least one of the average value, median value, mode value, cumulative value, maximum value, minimum value, rate of change, number of deviations from the reference range, or a preset state score.

[0288] For example, the device (200) can calculate the change in activity level by comparing the average activity level for 30 days before using the service with the average activity level for 30 days after using the service. Additionally, the device (200) can calculate the change in weight by comparing weight values ​​before and after using the service, or calculate the change in test results by comparing test values ​​before and after using the test kit.

[0289] The device (200) is not limited to a configuration that calculates the amount of change as a simple difference value. The device (200) can calculate the amount of change as an increase rate, decrease rate, score difference, grade change, whether it enters a normal range, whether it deviates from a normal range, or a status value indicating the direction of improvement, maintenance, or deterioration.

[0290] For example, if the average activity level in the reference interval data is 100 and the average activity level in the observation interval data is 120, the device (200) can calculate a change in lifestyle pattern indicating that the activity level has increased by 20. Additionally, if the test result moves from outside the reference range into the reference range, the device (200) can assign a state value of improvement direction to the change in the test result.

[0291] The device (200) can set different comparison items for each service type when calculating the amount of change. For example, for services related to nutrition management, weight, intake, and bowel status can be used as major comparison items, and for services related to activity management, activity level, walking time, and sleep amount can be used as major comparison items. For services related to health checkups or test kits, test result values ​​or health status scores can be used as major comparison items.

[0292] When calculating the amount of change in health status, the amount of change in test results, and the amount of change in lifestyle patterns, the device (200) may take into account differences in data collection intervals or measurement methods. For example, if reference interval data is collected on a daily basis and observation interval data is collected on a weekly basis, the device (200) may calculate the amount of change after converting them to the same comparison unit.

[0293] Through this, the device (200) can organize changes in the pet's condition before and after the use of the service into objective numerical values ​​or condition values. Accordingly, the device (200) can reflect in the feedback data for training an AI model how the pet's health condition, test results, or lifestyle patterns have changed after a specific service has been provided.

[0294] In step S307, the device (200) can assign a life cycle label according to pre-set life cycle label assignment criteria based on pet registration data, health history data, biometric data, lifestyle record data, changes in health status, changes in test results, and changes in lifestyle patterns.

[0295] Life cycle labels are classification values ​​assigned to distinguish the life cycle status of pets in feedback data used for AI model training. Life cycle labels are not limited to values ​​determined solely by the pet's age but can be determined by reflecting health history, biological condition, lifestyle, and changes before and after service usage.

[0296] The life cycle label assignment criteria are judgment criteria that the device (200) has pre-set for assigning a life cycle label. The life cycle label assignment criteria may be set to use at least one of species, breed, age, weight, disease history, amount of change in health status, amount of change in test results, amount of change in lifestyle pattern, and pre-stored standard values ​​for each life cycle.

[0297] For example, the device (200) can check the species, breed, age, and weight in the pet registration data and check for chronic diseases or a history of repeated medical treatments in the health history data. Additionally, the device (200) can check whether the activity level, intake, bowel movements, sleep duration, and weight changes in the biometric data and lifestyle record data fall within the standard range for each life cycle.

[0298] The device (200) can correct the life cycle label by using the amount of change in health status, the amount of change in test results, and the amount of change in lifestyle patterns together. For example, even if a pet is close to adulthood based on age, if a recent decrease in activity, change in weight, abnormal test results, or a history of disease is confirmed, the device (200) can assign a label indicating a condition requiring observation or an old age-related label according to the pre-set life cycle label assignment criteria.

[0299] The device (200) is not limited to a configuration that assigns a life cycle label as only one confirmed label. The device (200) may calculate a score value or probability value for multiple candidate labels and then assign the candidate label with the highest score value or a value greater than or equal to a preset standard as the life cycle label.

[0300] For example, the device (200) can calculate an adult candidate score, an old age candidate score, and a end-of-life care need candidate score for a specific pet. In this case, if the old age candidate score is above a preset standard and higher than other candidate scores, the device (200) can assign a life cycle label corresponding to old age to the training candidate data of the pet.

[0301] The device (200) can store the basis for assigning the life cycle label. For example, the device (200) can record which criterion was used for assigning the label, such as age criteria, disease history criteria, criteria for changes in test results, or criteria for changes in lifestyle patterns. Through this, the device (200) can verify the basis for assigning the life cycle label during the subsequent verification process of feedback data for training the AI ​​model.

[0302] Through this, the device (200) can assign a life cycle label by considering the pet's registration information, health history, biological condition, living condition, and changes before and after service use, rather than just a simple age classification. Accordingly, the feedback data for training the AI ​​model can include life cycle classification information close to the pet's actual condition, and can improve the accuracy of subsequent AI model training.

[0303] In step S308, the device (200) can assign a care result label by comparing the change in health status, the change in test results, and the change in lifestyle patterns with preset care effect judgment criteria.

[0304] Care result labels are classification values ​​assigned to distinguish which result type corresponds to a change in a pet's condition after using the service. Care result labels can be used as output or evaluation criteria to learn specific service delivery outcomes from feedback data used for AI model training.

[0305] The care effect determination criteria are criteria that the device (200) has pre-set to evaluate the amount of change before and after the use of the service. The care effect determination criteria may include at least one of the following: a standard for the amount of change in health status, a standard for the amount of change in test results, a standard for the amount of change in lifestyle patterns, a standard for the direction of change, a standard for the degree of change, and whether the standard range is satisfied.

[0306] The device (200) can compare the amount of change in health status with the care effect judgment criteria. For example, if the number of disease-related symptoms decreases after using the service, or if the abnormal behavior records entered through the guardian's terminal (100) decrease, the device (200) can determine whether the amount of change in health status corresponds to the direction of improvement. Conversely, if an increase in the number of medical consultations, records of worsening symptoms, or worsening of guardian observation records are confirmed, the device (200) can determine whether the amount of change in health status corresponds to the direction of deterioration.

[0307] The device (200) can also compare the amount of change in the test result with the care effect judgment criteria. For example, if the test kit result or the health check result approaches a preset normal range, the device (200) can determine that the amount of change in the test result satisfies the improvement criteria. Conversely, if the test value deviates from the normal range or the degree of deviation from the standard range increases, the device (200) can determine that the amount of change in the test result satisfies the deterioration criteria.

[0308] The device (200) can determine the care result using the amount of change in lifestyle patterns. For example, if the amount of activity recovers above a preset lower limit standard after using the service, the amount of intake stabilizes within a certain range, or the frequency of abnormal occurrences in the bowel state record decreases, the device (200) can determine that the amount of change in lifestyle patterns satisfies the improvement standard.

[0309] The device (200) is not limited to a configuration that assigns a care result label based on only one of the following items: a change in health status, a change in test results, and a change in lifestyle patterns. The device (200) may assign a care result label by comparing multiple change amounts together. For example, if the test results have improved but the activity level has decreased, the device (200) may determine the care result label based on a preset weight or priority criterion.

[0310] The care result label may be assigned with at least one label value among improvement, maintenance, deterioration, need for observation, insufficient data, or judgment pending. These label values ​​are examples, and the device (200) may use different label schemes depending on the service type, pet life cycle label, or data collection item.

[0311] For example, the device (200) may assign an improvement label to a case of service usage if, after using the nutrition management service, the amount of weight change stabilizes within a standard range and the amount of intake change satisfies the improvement criteria. Additionally, if test results are collected after health checkup guidance but there is insufficient lifestyle record data necessary to determine the change in condition, the device (200) may assign a data insufficient or judgment pending label.

[0312] The device (200) can store the basis for labeling along with the care result label. For example, the device (200) can record whether the criteria for the change in health status, the criteria for the change in test results, and the criteria for the change in lifestyle patterns have been satisfied. Through this, the device (200) can verify the reliability of the care result label during the subsequent process of generating training candidate data or training an AI model.

[0313] Through this, the device (200) can organize the results after service use not simply as whether use is completed, but as result labels based on changes in the pet's condition. Accordingly, the device (200) can reflect in the AI ​​model learning whether results such as improvement, maintenance, deterioration, or the need for observation occurred when a service was provided to a pet with a certain life cycle label.

[0314] In step S309, the device (200) can generate training candidate data by matching service usage feature data, life cycle labels, and care result labels with each other.

[0315] Training candidate data refers to data in the form of candidates that can be used as feedback data for training AI models. Training candidate data may include services provided to a specific pet, the guardian's response, service usage results, and correspondence relationships between life cycle labels and care outcome labels.

[0316] The device (200) can link a life cycle label and a care result label assigned to the same pet based on service identification information, whether the service was selected, the time of service use, and information on changes in the pet's condition after service use, which are included in the service usage characteristic data. At this time, the device (200) can use at least one of the pet identifier, service identification information, the time of service use, and the time of data generation as a matching criterion.

[0317] For example, the device (200) can generate a single training candidate data set of service usage characteristic data, life cycle label, and care result label when a health checkup service is provided to a pet assigned a life cycle label corresponding to old age, and the checkup results and changes in lifestyle records are collected after the guardian reserves the service through the guardian's terminal (100).

[0318] The device (200) can be configured to include both the state at the time of service provision and the result after service use in the training candidate data. For example, the training candidate data may include at least one of a life cycle label prior to service provision, service identification information, whether the service was selected, the time of service use, a care result label, and information on the basis for label assignment.

[0319] The device (200) can generate training candidate data for each service usage case when multiple services are provided to the same pet. For example, when a health checkup service and a nutrition management service are provided to the same pet at different times, the device (200) can generate different training candidate data based on each service identification information and the time of service usage.

[0320] The device (200) can indicate whether the data is complete when generating training candidate data. For example, if the service selection status and life cycle labels exist but the care result label is pending judgment, the device (200) can indicate the training candidate data as a restricted training target or a target requiring verification. This processing can reduce the unconditional inclusion of incomplete data in the training of the AI ​​model.

[0321] Through this, the device (200) can organize the service provision history, the guardian's reaction, the pet's life cycle status, and the results after using the service into a single learning unit. Accordingly, the device (200) can be structured to utilize for AI model training which care result label appeared when which service was provided to a pet with which life cycle label.

[0322] In step S310, the device (200) can classify the training candidate data by combination of life cycle labels and care result labels.

[0323] A combination of life cycle labels and care outcome labels is a classification unit for distinguishing which life cycle state and which care outcome correspond to the training candidate data. For example, the device (200) may use a unit combining a life cycle label and a care outcome label as a single combination, such as old age and improvement, old age and need for observation, adulthood and maintenance, end-of-life care need state and judgment pending.

[0324] The device (200) checks the life cycle labels and care result labels included in the training candidate data and can classify training candidate data having the same combination into the same group. At this time, the device (200) may use additional items such as service type, service identification information, or pet species as auxiliary criteria, but the basic classification criteria is the combination of the life cycle labels and care result labels.

[0325] For example, the device (200) may classify training candidate data with a life cycle label of old age and a care outcome label of improvement into a first combination group, and training candidate data with a life cycle label of old age and a care outcome label of deterioration into a second combination group. Additionally, training candidate data with a life cycle label of adulthood and a care outcome label of maintenance may be classified into a separate combination group.

[0326] The device (200) can calculate at least one of the number of training candidate data for each combination, the ratio, the distribution of service types, and the data generation period. For example, the device (200) can determine what percentage of the total training candidate data corresponds to the old age and improvement combination, and whether there is a sufficient number of data corresponding to the end-of-life care need state and improvement combination.

[0327] The device (200) can classify training candidate data, for which the care result label is judgment pending or data insufficient, into separate combination groups. This allows the device (200) to use the data separately from general training data during subsequent sample ratio correction or AI model training processes.

[0328] Through this, the device (200) can establish a structure that can determine whether the training candidate data is excessively biased toward a specific life cycle or a specific care outcome. Accordingly, in subsequent processing, differences in data distribution by combination can be determined, and if necessary, the sample ratio can be corrected to reduce the bias of the AI ​​model training data.

[0329] In step S311, the device (200) can correct the sample ratio of the data by combination if the difference in data distribution by combination of the life cycle label and the care result label satisfies a preset imbalance criterion.

[0330] The imbalance criterion is a criterion that the device (200) pre-sets to determine whether the distribution of the training candidate data by label combination is skewed to the extent that it is unsuitable for training an AI model. The imbalance criterion may include at least one of the number of data per combination, the ratio relative to the total data, the difference in ratio between the maximum combination and the minimum combination, the minimum sample size criterion, or the lack of a specific label combination.

[0331] For example, the device (200) may determine that a combination satisfies an imbalance criterion if the training candidate data corresponding to a combination of a specific life cycle label and a care outcome label is less than a preset minimum ratio of the total training candidate data. Additionally, if there is sufficient data for the combination of old age and improvement, while there is very little data for the combination of end-of-life care need and observation need, the device (200) may determine that the difference in data distribution by combination satisfies an imbalance criterion.

[0332] Correction of the sample ratio is a process of adjusting the usage ratio of training candidate data by combination to reduce the AI ​​model being trained with a bias toward a specific life cycle label or care outcome label. The device (200) can correct the sample ratio by increasing the sample weight of insufficient combinations, lowering the sample usage ratio of excessively many combinations, or resetting the training reflection ratio by combination.

[0333] The device (200) can adjust only the ratio or weight reflected in the learning without arbitrarily changing the original data. For example, the device (200) can assign a high learning weight to training candidate data with a small number of label combinations and a low learning weight to training candidate data with an excessively large number of label combinations. In this case, the original data stored in the pet life cycle learning database is maintained, and only the method of reflecting the training target data used for AI model training can be adjusted.

[0334] According to the embodiment, the device (200) may exclude some data of combinations with a large number of samples from the learning target, or set the device (200) to repeatedly use data of combinations with an insufficient number of samples. Additionally, the device (200) may mark combinations with insufficient data as combinations requiring separate collection, thereby managing the collection of additional data from the guardian's terminal (100) or an external service server thereafter.

[0335] For example, if the device (200) has most of the training candidate data for the adulthood and maintenance combination and lacks training candidate data for the old age and deterioration combination or the end-of-life care need state and observation need combination, it can increase the training reflection weight of the lacking combination. Through this, the AI ​​model can learn not only common combinations with a lot of data but also life cycle states and care outcomes with relatively little data.

[0336] The device (200) can store the sample ratio correction results together with the training candidate data. The sample ratio correction results may include at least one of the number of data per combination, the ratio per combination, the reflection ratio before correction, the reflection ratio after correction, the sample weight, and the reason for correction.

[0337] Through this, the device (200) can reduce the reflection of training candidate data biased toward specific life cycle labels or care outcome labels directly into the training of the AI ​​model. Accordingly, it can compensate for the quality of learning regarding conditions requiring care during old age, end-of-life care, or specific care outcomes, where relatively little data may be collected, and can increase the balance and reliability of feedback data for training the AI ​​model.

[0338] In step S312, the device (200) can generate feedback data for training an AI model based on training candidate data that reflects the correction result of the sample ratio.

[0339] Training candidate data reflecting the correction result of the sample ratio is training candidate data in which data imbalance by combination of life cycle labels and care result labels is corrected. The device (200) can generate feedback data for training an AI model by reflecting at least one of the sample weight by combination, the training reflection ratio, whether to exclude, whether to use repeatedly, or whether verification is required.

[0340] The device (200) can organize service usage feature data, life cycle labels, and care result labels included in the training candidate data into a structure that can be used for training an AI model. For example, the device (200) can organize service identification information, whether the service was selected, the time of service use, life cycle labels, care result labels, and sample weights into a single training record.

[0341] When the device (200) generates feedback data for training an AI model, it can distinguish between input items used for training and learning objectives or validation items. For example, input items may include feature values ​​derived from service usage feature data and pet management data, and learning objectives or validation items may include life cycle labels, care result labels, or a combination thereof.

[0342] The device (200) may include the sample ratio correction results together with the feedback data for training the AI ​​model. For example, the device (200) may assign a first sample weight to training candidate data corresponding to the combination of old age and improvement, and assign a second sample weight to training candidate data corresponding to the combination of end-of-life care need and observation need. In this case, the first sample weight and the second sample weight may be set differently depending on the number of data or the degree of imbalance for each combination.

[0343] The device (200) can determine data among the training candidate data that satisfies the data validity criteria as feedback data for training an AI model. For example, training candidate data in which the pet identifier, service identification information, life cycle label, and care result label are all verified and are not excluded by the exclusion criteria can be set as trainable data. Conversely, data in which the care result label is pending judgment or the observation period is insufficient can be classified as data requiring verification or limited training data.

[0344] The device (200) can generate feedback data for AI model training in a format that can be stored in a pet lifecycle learning database. For example, the device (200) can generate a data structure including training record identification information, pet identifier, service identification information, lifecycle label, care result label, sample weight, time of generation, and data validity status.

[0345] Through this, the device (200) can generate training data in which imbalances are corrected for combinations of life cycle labels and care result labels, rather than simply accumulating service usage results. Accordingly, the AI ​​model can learn not only general cases where a large amount of data is collected, but also life cycle states or care results where data is relatively scarce, and can improve the accuracy of service matching according to the condition of the pet.

[0346] As a result, the device (200) can label changes in the pet's condition before and after service use and correct data imbalances by combination to generate feedback data suitable for subsequent learning of the AI ​​model.

[0347] FIG. 4 is a flowchart illustrating the process of generating life cycle measurement management information according to one embodiment.

[0348] Specifically, the device (200) may generate life cycle measurement management information after calculating life cycle status data and care need data, and before generating service matching data.

[0349] Lifecycle measurement and management information is data indicating which items require more precise measurement or management based on the pet's current lifecycle state. This information may include precision measurement condition data, measurement validity data, measurement deviation data, data on items requiring improvement, and data on candidates for improvement services.

[0350] The device (200) can determine the current level of care required of a pet using life cycle status data and care need data, and select measurement items with a high impact among the levels of care required. For example, if the activity level management need value and the weight change management need value are calculated to be high in a pet corresponding to the senior stage, the device (200) can set the activity level and weight change as the main targets for life cycle measurement management.

[0351] Improvement service candidate data is data included in life cycle measurement management information, representing service candidates deemed necessary for the pet based on measurement results. Improvement service candidate data may correspond to at least one of health checkup guidance, test kit usage guidance, activity management guidance, nutrition management guidance, daily record input guidance, or regular monitoring guidance.

[0352] The device (200) can set improvement service candidate data as service candidate data used to generate service matching data according to preset candidate setting criteria. The candidate setting criteria may include at least one of whether the criteria for measurement validity data are met, the magnitude of measurement deviation data, the type of item data requiring improvement, the level of care need data, whether guardian condition data is met, and the conditions for service provision.

[0353] For example, if the device (200) determines that the weight change in the key factor measurement results deviates significantly from the statistical standards of the same species and same life cycle, and that the deviation in weight change is large even when compared with similar population data, it can generate nutritional management services or health checkup services as candidate data for improvement services. Subsequently, the device (200) can set the said candidate data for improvement services as candidate data for service matching data generation.

[0354] Through this, the device (200) can configure service candidates by selecting items that actually need to be measured and managed in the pet's current life cycle once more, rather than matching services immediately using only life cycle status data and care need data.

[0355] Accordingly, service matching data can be generated based on candidate data that reflects the impact of measurement items, measurement deviations, and the need for improvement, rather than simply the results of lifecycle classification.

[0356] Referring to FIG. 4, first, in step S401, the device (200) can extract similar population data by comparing the pet registration data, life cycle status data, and health history data with the individual history data stored in the pet life cycle learning database according to preset similar individual selection criteria.

[0357] Individual history data is past history data for each pet stored in a pet life cycle learning database. Individual history data may include at least one of the pet's species, breed, age, weight, gender, life cycle status, health history, biometric data, living record data, service usage history data, measurement data, and care result labels.

[0358] The criteria for selecting similar individuals are comparison criteria that the device (200) has pre-set to select individuals having characteristics similar to the pet currently being analyzed. The criteria for selecting similar individuals may include at least one of a species match criterion, a breed similarity criterion, an age range criterion, a weight range criterion, a life cycle status match criterion, a disease history similarity criterion, a health checkup result similarity criterion, a lifestyle pattern similarity criterion, and a data retention amount criterion.

[0359] The device (200) can extract at least one of species, breed, age, weight, and gender from pet registration data and determine from life cycle status data which state the current pet corresponds to: growth stage, adult stage, old age, end-of-life care required state, or observation required state. Additionally, the device (200) can determine at least one of disease history, medical treatment history, health checkup results, test results, medication history, or recurring abnormality items from health history data.

[0360] The device (200) can compare the data confirmed in this way with individual history data stored in the pet life cycle learning database. For example, if the pet being analyzed is a small dog aged 12 years, and recent weight changes and a decrease in activity levels are confirmed, and life cycle status data corresponding to the elderly stage is produced, the device (200) can set individual history data having the same or similar species, similar weight range, similar age group, and elderly life cycle status as a priority comparison target.

[0361] The device (200) is not limited to a configuration that simply selects individual history data satisfying similar individual selection criteria based on only one condition. The device (200) can calculate similarity for multiple comparison items and calculate an individual similarity score by combining the calculated similarities. For example, the device (200) can score the species match, breed similarity, age difference, weight difference, life cycle status match, and disease history similarity, respectively, and then calculate an overall similarity score according to a preset weight.

[0362] The device (200) can extract individual history data as similar population data where the overall similarity score is greater than or equal to a preset similarity standard. Additionally, the device (200) can be configured to extract a certain number of individual history data in order of highest similarity score, or to extract only those individuals that satisfy specific essential conditions. For example, the same species and the same life cycle state can be set as essential conditions, and the similarity of breed, weight, disease history, and life records can be used as scoring criteria.

[0363] Similar population data is data regarding a group of pets that has a history similar to the pet currently under analysis. Similar population data may include at least one of identification information, similarity score, life cycle status, health history, biometric data, living record data, measurement data, service usage history data, and care outcome data of similar individuals.

[0364] When extracting similar population data, the device (200) may exclude individual history data with low data reliability or insufficient observation period. For example, individual history data in which health history data is insufficient, measurement data has not been collected for a certain period or longer, or identification information is unclear may be excluded from the similar population data.

[0365] Through this, the device (200) can compare the pet subject to analysis with a population similar in species, breed, age, weight, life cycle status, and health history, rather than simply comparing it with the entire pet population. Accordingly, in subsequent processing, deviations in key factor measurement data, items requiring improvement, and candidate data for improvement services can be generated based on more objective comparison criteria.

[0366] In step S402, the device (200) can calculate impact data for each measurement item by applying a preset impact analysis criterion to at least one of the activity level management need value, intake level management need value, bowel state management need value, sleep level management need value, and weight change management need value included in the care need data.

[0367] Impact data by measurement item is data indicating which measurement item has a relatively large influence on management results or the judgment of changes in condition during the pet's current life cycle state. The impact data by measurement item may include an impact score, impact grade, priority value, or direction of influence information corresponding to at least one of activity level, intake, bowel status, sleep amount, and weight change.

[0368] The impact analysis criteria are criteria pre-set by the device (200) to compare each care need value included in the care need data and to calculate the importance of each measurement item. The impact analysis criteria may include at least one of the magnitude of the care need value, the correlation with life cycle status data, the correlation with health history data, the degree of recurrence in similar population data, the amount of recent change, whether it deviates from the standard range, and data reliability.

[0369] The device (200) can check the activity level management need value, intake level management need value, bowel state management need value, sleep level management need value, and weight change management need value included in the care need data. For example, the care need data may consist of numerical values ​​such as an activity level management need value of 85 points, an intake level management need value of 60 points, a bowel state management need value of 75 points, a sleep level management need value of 50 points, and a weight change management need value of 90 points.

[0370] The device (200) can apply weights based on life cycle status data to each management need value. For example, if the life cycle status data corresponds to the elderly stage, the device (200) can set weights regarding activity level, sleep amount, and weight change relatively high. If the life cycle status data corresponds to the growth stage, the device (200) can set weights regarding intake amount and weight change relatively high.

[0371] The device (200) can reflect the correlation with health history data in the impact analysis criteria. For example, if a history of digestive-related diseases is confirmed in the health history data, the device (200) can calculate a high impact value for the need to manage bowel status and the need to manage intake. If a history of joint disease or reduced activity is confirmed, the device (200) can calculate a high impact value for the need to manage activity level.

[0372] The device (200) can reflect the repeatability of measurement items identified in similar population data in the calculation of impact. For example, if weight changes and reduced activity levels repeatedly appear in pets with the same life cycle and similar health history in similar population data, the device (200) can calculate high impact data for the weight change management need value and the activity level management need value.

[0373] The device (200) is not limited to a configuration that calculates impact data for each measurement item simply based on the magnitude of the care need value. The device (200) can calculate a comprehensive impact score by applying a life cycle weight, a health history weight, similar population relevance, and recent change trends to the care need value.

[0374] For example, if the activity level management need value is 80 points, the elderly life cycle weight is applied, and a decrease in activity level is repeatedly confirmed in similar population data, the device (200) can produce high impact data for the activity level item. Conversely, even if the intake level management need value is at a certain level, if recent changes are small and the correlation with health history data is low, the impact data for the intake level item can be produced relatively low.

[0375] The device (200) can convert impact data by measurement item into impact score as well as impact grade or priority information. For example, the device (200) can set an item with an impact score greater than or equal to a preset first standard as a high grade, set an item with an impact score greater than or equal to a second standard but less than the first standard as a medium grade, and set an item with an impact score less than the second standard as a low grade.

[0376] The device (200) can store the calculated impact data for each measurement item by correlating it with the pet identifier, life cycle status data, and care need data. Through this, the device (200) can determine which measurement item should be prioritized for subsequent precision measurement and service candidate setting in the current life cycle status of a specific pet.

[0377] Through this, the device (200) does not treat various care need values ​​included in the care need data equally, but can calculate the impact of each measurement item based on the pet's life cycle status, health history, and the history of similar populations. Accordingly, in subsequent processing, key factor data is not selected arbitrarily, but can be selected based on measurement items that are highly relevant to the current state change of the pet.

[0378] In step S403, the device (200) can extract items whose impact data per measurement item is greater than or equal to a preset impact standard as key factor data.

[0379] Key factor data is data representing measurement items requiring precise measurement or priority management in the current life cycle state of a pet. Key factor data may correspond to at least one of activity level, intake, bowel status, sleep amount, and weight change, and may include at least one of an impact score, impact grade, priority information, and basis for extraction information for each item.

[0380] The impact criterion is a criterion pre-set by the device (200) to determine the target to be extracted as key factor data among the impact data for each measurement item. The impact criterion may include at least one of a lower limit value of the impact score, an impact grade criterion, a top ranking criterion, a standard value for each item by life cycle, or a correction criterion by health history.

[0381] The device (200) can compare the impact data for each measurement item corresponding to activity level, intake, bowel movement status, sleep amount, and weight change with an impact standard, respectively. For example, if the impact standard is set to 70 points or higher, the device (200) can extract items with an activity level impact score of 85 points and a weight change impact score of 90 points as core factor data. On the other hand, if the sleep amount impact score is 45 points, the corresponding item can be excluded from the core factor data.

[0382] The device (200) is not limited to a configuration that extracts core factor data based solely on impact scores. The device (200) may apply different impact criteria depending on the life cycle status data. For example, for pets in the elderly stage, the impact criteria regarding activity level, sleep amount, and weight change may be set relatively low to increase the likelihood that these items will be extracted as core factor data. For pets in the growth stage, the impact criteria regarding intake amount and weight change may be applied first.

[0383] The device (200) may adjust the impact criteria based on health history data. For example, if a history of digestive-related diseases is identified in the health history data, the device (200) may lower the impact criteria for intake and bowel status items or apply a correction weight to those items. If a history of joint disease or reduced activity is identified, the impact criteria may be adjusted so that the activity amount item is extracted as a key factor data.

[0384] If multiple items satisfy the impact criteria, the device (200) can extract each item as core factor data. In this case, the device (200) can set the priority of the core factor data according to the impact score or impact grade. For example, if the impact score of weight change is the highest and the impact score of activity level is the next highest, the device (200) can set weight change as the first core factor and activity level as the second core factor.

[0385] The device (200) can set essential life cycle measurement items as key factor data even if there are no items satisfying the impact criteria. For example, even if all impact scores are below the criteria, if weight change or activity level is set as an essential measurement item for an elderly pet, the device (200) can extract that item as the minimum key factor data. This configuration is intended to reduce the omission of necessary management items due to a lack of data or temporary omission of measurements.

[0386] The device (200) can store the reason for extraction together with the key factor data when extracting it. For example, the device (200) can distinguish and store whether a specific item was extracted because it exceeded an impact standard, because it was extracted because it was an essential item for each life cycle, or because it satisfied a correction standard based on health history.

[0387] The device (200) can store extracted key factor data by correlating it with pet identifiers, life cycle status data, care need data, and impact data by measurement item. Through this, the device (200) can set measurement cycles, measurement methods, and measurement precision conditions for each key factor data in subsequent processing.

[0388] Through this, the device (200) can select items with high actual management impact among activity level, intake, defecation status, sleep amount, and weight change as key factor data. Accordingly, subsequent precise measurement condition data is not generated uniformly for all items, but can be generated based on items that are substantially relevant to the pet's life cycle status and care needs.

[0389] In step S404, the device (200) can generate precision measurement condition data including a measurement period, measurement method, and measurement precision conditions corresponding to the core factor data.

[0390] Precision measurement condition data is data that determines what criteria the device (200) will use to measure items corresponding to key factor data. Precision measurement condition data may include at least one of a measurement cycle per key factor, a measurement method, a measurement precision condition, a measurement time zone, a measurement duration, a measurement data format, and a re-measurement condition in case of measurement failure.

[0391] The measurement period is a condition indicating at what interval items corresponding to key factor data will be measured. For example, the device (200) can be set to measure activity levels on a daily or hourly basis, weight changes on a weekly or monthly basis, and bowel movements can be set to be entered at the time of occurrence.

[0392] The measurement method is a condition indicating how items corresponding to core factor data will be collected. For example, the device (200) can be configured to collect activity levels through a wearable device, a sensor device linked to the guardian's terminal (100), or direct input from the guardian. Additionally, intake, bowel movements, and sleep levels can be configured to be collected through input values ​​via the guardian's terminal (100), measurements from an external device, or input values ​​from lifestyle records within the platform.

[0393] The measurement precision condition is a condition indicating how finely or with what level of confidence the items corresponding to the key factor data will be measured. For example, the device (200) can be set to input weight changes down to the decimal point, or to collect activity levels by time period along with the total daily activity amount. Additionally, regarding bowel status, it can be set to input not only whether it simply occurred, but also the number of bowel movements, a status classification value, and whether there were any abnormalities.

[0394] The device (200) can generate different precision measurement condition data depending on the type of key factor data. For example, if the key factor data is activity level, the device (200) can generate precision measurement condition data to measure at least one of daily activity level, activity level by time of day, walking time, or distance traveled. If the key factor data is intake, the device (200) can generate precision measurement condition data to measure at least one of feed amount, water intake, feeding frequency, or appetite change input value.

[0395] The device (200) can adjust the measurement cycle and measurement precision conditions by reflecting life cycle status data and care need data. For example, if the life cycle status data is in the elderly stage and the weight change management need value is high, the device (200) can set the measurement cycle for weight change to be short and generate precision measurement condition data to provide a request for regular weight input to the guardian's terminal (100).

[0396] The device (200) can set different measurement conditions according to the priority of the impact data or key factor data for each measurement item. For example, for weight change set as the first key factor, a measurement condition of once a week can be set, and for activity amount set as the second key factor, daily automatic measurement or daily input conditions can be set.

[0397] The device (200) can generate precise measurement condition data by considering the data reliability of each measurement method. For example, for items where external device measurement values ​​exist, automatic collection can be set as the priority measurement method, and for items where automatic collection is difficult, an input method through the guardian's terminal (100) can be set. In addition, for items directly input by the guardian, notification conditions or input format restriction conditions to prevent input omission can be set together.

[0398] For example, if activity level and weight change are extracted as key factor data, the device (200) can generate precise measurement condition data including conditions for automatic daily collection of activity level, generation of summary values ​​by time period, and calculation of deviation from standard activity level. Additionally, for weight change, it can generate precise measurement condition data including conditions for input once a week, input of measurement values ​​to the first decimal place, and calculation of the rate of change relative to the previous measurement value.

[0399] The device (200) can store the generated precision measurement condition data in correspondence with the pet identifier, key factor data, life cycle status data, and care need data. Through this, the device (200) can apply the same measurement conditions when collecting key factor measurement data from the guardian's terminal (100), an external device, or an external service server thereafter.

[0400] Through this, the device (200) can not only simply display items corresponding to key factor data as management targets, but also convert them into specific measurement conditions including measurement cycles, measurement methods, and measurement precision conditions. Accordingly, the format and quality of key factor measurement data collected in subsequent processing can be maintained consistently, and measurement validity data and measurement deviation data can be calculated more reliably.

[0401] In step S405, the device (200) can collect key factor measurement data corresponding to key factor data according to precision measurement condition data.

[0402] Key factor measurement data is an actual measured value or recorded value collected for a measurement item extracted as key factor data, according to the measurement cycle, measurement method, and measurement precision conditions determined in the precision measurement condition data. The key factor measurement data may include at least one of activity level measurement data, intake level measurement data, bowel status measurement data, sleep level measurement data, and body weight change measurement data.

[0403] The device (200) can determine the timing of the collection of core factor measurement data based on the measurement cycle included in the precision measurement condition data. For example, if the measurement cycle of activity amount in the precision measurement condition data is set to daily, the device (200) can collect daily activity amount data. If the measurement cycle of weight change is set to once a week, the device (200) can provide a request for weight input once a week to the guardian's terminal (100) and collect the entered weight value as core factor measurement data.

[0404] The device (200) can collect key factor measurement data from at least one of a guardian's terminal (100), an external device linked to the guardian's terminal (100), an external service server, or a life record input screen within a platform, according to the measurement method included in the precision measurement condition data. For example, activity level can be collected from a wearable device or an activity level sensor, and intake amount and bowel status can be collected through a direct input method via the guardian's terminal (100).

[0405] The device (200) can control the format and details of the collected data by checking the measurement precision conditions included in the precision measurement condition data. For example, if the device (200) is set to input weight changes up to the first decimal place, it can receive a weight value corresponding to that format from the guardian's terminal (100). If the device (200) is set to input the number of bowel movements, the condition classification value, and whether there is an abnormality together regarding the bowel movement status, it can provide a screen containing the corresponding input items to the guardian's terminal (100).

[0406] If the core factor data is activity level, the device (200) can collect at least one of the total activity level, activity level by time of day, walking time, distance traveled, and whether activity level has decreased as core factor measurement data. If the core factor data is intake, the device (200) can collect at least one of the feed amount, water intake, feeding frequency, change in appetite, and feeding time as core factor measurement data.

[0407] If the core factor data is a defecation state, the device (200) can collect at least one of the following as core factor measurement data: number of defecations, time of defecation, classification value of the defecation state, whether there is an abnormality in defecation, and a guardian observation note. If the core factor data is a amount of sleep, the device (200) can collect total sleep time, time of sleep start, time of sleep end, number of night awakenings, or information on changes in sleep patterns. If the core factor data is a change in weight, the device (200) can collect at least one of the following: measured weight value, time of measurement, amount of change compared to the previous measurement value, and rate of change compared to the reference weight.

[0408] When collecting key factor measurement data, the device (200) can correspond information such as a pet identifier, measurement item, measurement time, data source, and measurement method. Through this, the device (200) can distinguish and manage key factor measurement data for each pet even when the same guardian has registered multiple pets.

[0409] The device (200) can verify whether the collected key factor measurement data corresponds to the precision measurement condition data. For example, data entered later than the measurement cycle, data that does not satisfy the measurement precision condition, data with missing measurement items, or data with measurement values ​​outside the allowable range may be classified as subjects for correction, subjects for re-entry requests, or subjects for validation.

[0410] For example, if the core factor data is weight change and the precision measurement condition data requires weight measurement once a week, the device (200) can provide a weight input notification to the guardian's terminal (100) and collect the weight value entered by the guardian and the time of measurement. If the entered weight value shows an abnormally large difference compared to the previous measurement value, the device (200) can manage the data as a subject for reconfirmation request without immediately reflecting the data in the analysis.

[0411] Additionally, if the core factor data is activity level and the measurement method is set to an external device linkage method, the device (200) can receive daily activity level data from the guardian's terminal (100) or an external device. The device (200) can organize the received activity level data into summary values ​​by time period or deviation values ​​relative to the standard activity level according to the precise measurement condition data.

[0412] Through this, the device (200) can collect actual measurement data corresponding to key factor data in accordance with precise measurement conditions. Accordingly, it can reliably secure basic data for comparison with statistical reference data of the same species and same life cycle in subsequent processing, or for comparison with corresponding measurement data included in similar population data.

[0413] In step S406, the device (200) can calculate measurement validity data by comparing the key factor measurement data with statistical reference data of a group of pets belonging to the same species and the same life cycle.

[0414] Statistical reference data is data representing the distribution of measurements of a pet population stored in a pet life cycle learning database or an external reference database. Statistical reference data may include at least one of the mean, median, standard deviation, upper reference value, lower reference value, quantile reference value, normal range information, and abnormal range information of a pet population belonging to the same species and the same life cycle.

[0415] A group of pets belonging to the same species and same life cycle refers to a group of pets that is of the same species as the pet being analyzed and whose life cycle status data is identical or corresponds to a range. For example, if the pet being analyzed is a dog in the elderly stage, the device (200) may use statistical reference data regarding activity level, intake, defecation status, sleep amount, or weight change of the group of dogs in the elderly stage as a comparison standard.

[0416] Measurement validity data is data indicating whether key factor measurement data can be regarded as valid measurements when compared to statistical standards of the same species and same life cycle group. Measurement validity data may include at least one of the following: validity, satisfaction of the standard range, degree of deviation from the standard range, statistical deviation value, reliability score, necessity of re-measurement, and reason for determining validity.

[0417] The device (200) can compare the measurement values ​​for each measurement item included in the key factor measurement data with statistical reference data. For example, if the key factor data is activity level, the device (200) can compare the daily activity level measurement of the pet being analyzed with the average activity level, lower limit reference value, or quantile reference value of the same species and same life cycle group.

[0418] When the core factor data is weight change, the device (200) can compare the measured weight value or weight change rate with the weight change statistical reference data of the same species and same life cycle group. For example, if the recent weight loss rate of the pet being analyzed exceeds the statistically acceptable range of the same species and same life cycle group, the device (200) can generate measurement validity data including reference range deviation information for the corresponding weight change measurement value.

[0419] When the core factor data is a defecation state, the device (200) can compare the number of defecations, the classification value of the defecation state, or the frequency of abnormal defecation occurrence with statistical reference data of the same species and same life cycle group. For example, if the frequency of abnormal defecation occurrence is higher than the upper limit reference value of the same species and same life cycle group, the device (200) can calculate measurement validity data indicating that the defecation state measurement data corresponds to an abnormal range.

[0420] The device (200) is not limited to a configuration that determines only whether the measurement value is within the normal range of statistical reference data. The device (200) can determine how much the measurement value deviates from the reference range, whether it deviates repeatedly, and whether the degree of deviation has increased over a recent period. For example, if the activity level measurement value is lower than the lower limit reference value and the same condition is repeated over several measurement cycles, the device (200) can include continuous deviation information in the validity determination result of the activity level measurement data.

[0421] The device (200) can also reflect the method of collecting key factor measurement data in determining measurement validity. For example, activity data automatically collected from an external device may have a high reliability score for the data source, and data directly entered through the guardian's terminal (100) may have a reliability score that varies depending on whether input is missing, whether the input format is satisfied, or whether reconfirmation is required.

[0422] When calculating measurement validity data, the device (200) can check whether the measurement cycle is satisfied and whether the measurement precision condition is satisfied. For example, if weight measurement is set once a week in the precision measurement condition data but actual measurement data is collected only once a month, the device (200) can calculate the validity of the weight change measurement data as low or indicate that re-measurement is required.

[0423] For example, if the core factor data is weight change and the recent 4-week weight loss rate of the pet being analyzed falls outside the sub-standard range of the same species and same life cycle group, the device (200) can calculate values ​​indicating the deviation from the standard range, the degree of deviation, whether re-measurement is necessary, and the possibility of linking to a health checkup as measurement validity data.

[0424] Additionally, if the core factor data is activity level and the daily activity level is lower than the average value of the same species and same life cycle group, but the measurement data is collected continuously for a certain period according to precise measurement conditions, the device (200) can produce measurement validity data that the data collection reliability is high and the activity level is lower than the standard for the activity level measurement data.

[0425] The device (200) can store the calculated measurement validity data by correlating it with identification information of pet identifiers, key factor data, key factor measurement data, and statistical reference data. Through this, the device (200) can verify whether the key factor measurement data is a reliable comparison target when generating measurement deviation data and improvement-needed item data in subsequent processing.

[0426] Through this, the device (200) can determine the validity of the measurement value and the degree of deviation from the standard by comparing it with statistical standards of the same species and same life cycle group, rather than using the core factor measurement data merely as a collected value. Accordingly, data on items requiring subsequent improvement and candidate data for improvement services can be generated more objectively by reflecting the results of comparison with statistical standards, rather than temporary fluctuations in individual measurements.

[0427] In step S407, the device (200) can generate measurement deviation data and improvement need item data by comparing key factor measurement data with corresponding measurement data included in similar population data.

[0428] Corresponding measurement data refers to data included in similar population data that corresponds to measurement items such as the key factor data of the pet under analysis. For example, if the key factor data of the pet under analysis is activity level, the corresponding measurement data may include activity level measurements, average activity levels, activity level change rates, or activity level distribution values ​​included in the similar population data.

[0429] Here, statistical reference data for a group of pets belonging to the same species and life cycle may serve as a criterion for determining whether the key factor measurement data falls within a general reference range, and corresponding measurement data included in similar population data may serve as a criterion for determining the relative deviation from a population with a health history or life history similar to that of the pet under analysis.

[0430] Measurement deviation data is data indicating the extent to which the measurement data of the key factors of the pet under analysis differs from the corresponding measurement data of similar population data. Measurement deviation data may include at least one of a difference value, a difference in the rate of change, a ratio relative to a reference, a rank value, a quantile value, a standardized deviation value, the direction of deviation, and whether the deviation persists.

[0431] Data on items requiring improvement is data representing measurement items deemed to require management or service linkage based on measurement deviation data. The data on items requiring improvement may include at least one of items requiring improvement in activity level, items requiring improvement in intake, items requiring improvement in bowel status, items requiring improvement in sleep amount, and items requiring improvement in weight change.

[0432] Before comparing the core factor measurement data and the corresponding measurement data, the device (200) can verify whether the measurement items, measurement period, measurement unit, and measurement method are comparable to each other. For example, if the activity level of the pet subject to analysis is collected on a daily basis and the activity level of similar population data is stored as a weekly average value, the device (200) can perform the comparison after converting to the same comparison unit.

[0433] The device (200) can calculate a representative value of corresponding measurement data included in similar population data. The representative value may be at least one of an average value, a median value, a mode value, an upper reference value, a lower reference value, a quantile reference value, or a reference range by life cycle. The device (200) can generate measurement deviation data by comparing the key factor measurement data with the corresponding representative value or reference range.

[0434] For example, if the daily activity level of the pet being analyzed is lower than the median daily activity level of similar population data by more than a certain percentage, the device (200) can generate measurement deviation data in the direction of insufficient activity. Additionally, if the weight loss rate of the pet being analyzed exceeds the allowable range of the weight change rate included in the similar population data, the device (200) can generate measurement deviation data including the magnitude of the deviation and the direction of the deviation for the weight change item.

[0435] The device (200) is not limited to a configuration that generates measurement deviation data using only the difference at a single point in time. The device (200) can determine whether the difference between the key factor measurement data and the corresponding measurement data is repeated, increasing, or decreasing over multiple measurement cycles. For example, cases where the activity deviation occurs once and cases where it occurs repeatedly for three weeks or more can be distinguished as different deviation states.

[0436] The device (200) can generate data on items requiring improvement by reflecting measurement validity data together. For example, if the core factor measurement data is determined to be a valid measurement value when compared with statistical reference data, and at the same time the deviation is greater than or equal to a preset deviation threshold when compared with the corresponding measurement data of similar population data, the device (200) can generate the corresponding measurement item as data on items requiring improvement.

[0437] The device (200) can generate data on items requiring improvement based on whether the measured deviation data is above a preset deviation standard, whether the direction of the deviation is a direction requiring management attention, whether the deviation persists for a certain period of time or longer, and whether the care need value of the item is high. For example, if the activity level of an elderly pet is consistently lower than that of a similar population and the activity level management need value is also high, the device (200) can set the activity level as an item requiring improvement.

[0438] The device (200) can generate measurement deviation data and improvement-needing item data for each of the multiple key factor data. For example, if the key factor data are activity level and weight change, the device (200) can generate first measurement deviation data by comparing the activity level measurement data with activity level corresponding measurement data of a similar population, and generate second measurement deviation data by comparing the weight change measurement data with weight change corresponding measurement data of a similar population.

[0439] The device (200) can assign an improvement priority to the data of items requiring improvement. The improvement priority can be determined based on at least one of the size of the measurement deviation data, the duration of the deviation, the reliability of the measurement validity data, the level of the care need data, and the life cycle status data. For example, if the weight change deviation is large, the measurement validity is high, and it corresponds to an old age life cycle status, the device (200) can set the improvement priority of the weight change item high.

[0440] The device (200) can store the generated measurement deviation data and improvement-needing item data by correlating them with pet identifiers, key factor data, similar population data, and measurement validity data. Through this, the device (200) can verify which measurement item was determined to be an improvement-needing item based on what criteria when generating improvement service candidate data in subsequent processing.

[0441] Through this, the device (200) can compare the measurement value of the pet subject to analysis with corresponding measurement data of a similar population having similar life cycle status and health history, rather than comparing it with a simple overall average. Accordingly, it can more precisely identify items requiring actual improvement while reflecting differences based on the pet's species, age, weight, life cycle status, and health history.

[0442] In step S408, the device (200) can generate improvement service candidate data based on measurement validity data, measurement deviation data, and improvement need item data.

[0443] Improvement service candidate data is data representing candidate services that may be subject to service matching for improving the condition of a pet or for additional care based on the results of measuring key factors. Improvement service candidate data may include at least one of a service type, service identification information, items to be improved, grounds for recommendation, priority of provision, level of need for provision, and guidance items to be provided to the guardian's terminal (100).

[0444] The device (200) can determine whether the core factor measurement data is reliable data that can be used to generate service candidates by using measurement validity data. For example, if the weight change measurement data satisfies the measurement cycle and measurement precision conditions of the precision measurement condition data and is determined to be comparable to statistical reference data of the same species and same life cycle group, the device (200) can use the measurement value to generate improvement service candidate data.

[0445] The device (200) can use measurement deviation data to determine which items the pet subject to analysis shows a difference from similar population data. For example, if the activity level measurement data is consistently lower than the corresponding measurement data included in the similar population data, and the measurement deviation data is greater than or equal to a preset deviation threshold, the device (200) can generate candidate data for improvement services related to activity level management.

[0446] The device (200) can determine target items of service candidates using data on items requiring improvement. For example, if the data on items requiring improvement includes items requiring improvement regarding weight change, the device (200) can generate at least one of nutrition management guidance, health checkup guidance, test kit usage guidance, or regular weight input guidance as improvement service candidate data. If the data on items requiring improvement includes items requiring improvement regarding bowel condition, the device (200) can generate at least one of intake recording guidance, bowel condition recording guidance, test kit usage guidance, or health counseling linkage guidance as improvement service candidate data.

[0447] When generating service candidates, the device (200) can verify the correspondence between items requiring improvement and service types according to pre-stored service candidate mapping criteria. The service candidate mapping criteria are criteria indicating which service types correspond to items requiring improvement, such as activity level, intake, bowel movements, sleep amount, and weight change. For example, items requiring improvement in activity level may correspond to an activity management service, a walking record management service, or a lifestyle pattern monitoring service, and items requiring improvement in weight change may correspond to a nutrition management service, a health checkup service, or a weight record management service.

[0448] The device (200) can calculate the required level of provision of improvement service candidate data by reflecting together the reliability of measurement validity data, the magnitude of measurement deviation data, and the priority of improvement item data. For example, if the reliability of measurement validity data is high, the magnitude of deviation of measurement deviation data is large, and the priority of improvement item data is high, the device (200) can set the required level of provision of improvement service candidate data corresponding to the item to be high.

[0449] The device (200) is not limited to a configuration that generates improvement service candidate data as only one service. Multiple service candidates may be generated for a single item requiring improvement, and a single service candidate may correspond to multiple items requiring improvement. For example, if both weight change and activity level are determined to be items requiring improvement, the device (200) may generate improvement service candidate data that includes health checkup guidance, nutrition management guidance, and activity management guidance together.

[0450] After generating improvement service candidate data, the device (200) may store at least one of a service type, service identification information, cost condition, available area, provision method condition, and notification condition corresponding to the improvement service candidate data so that it can be compared with guardian condition data during the subsequent service matching data generation process. In this case, the guardian condition data is not limited to being an essential criterion for generating the improvement service candidate data itself, but can be used as a comparison criterion to determine the availability, priority of provision, or notification condition of the service candidate during the subsequent service matching data generation process.

[0451] The device (200) may include recommendation basis information in the improvement service candidate data. The recommendation basis information is information indicating which measurement validity data, measurement deviation data, and improvement need item data the corresponding service candidate was generated based on. For example, the device (200) may include basis information such as “sub-range of activity level compared to the same species and same life cycle group,” “above the standard for weight loss deviation compared to similar populations,” and “increased frequency of abnormal bowel condition occurrences” in the improvement service candidate data.

[0452] For example, if the validity of activity measurement data for a pet in its senior years is high, the deviation in activity compared to similar population data is large, and the activity level is set as an item requiring improvement, the device (200) can generate activity management guidance or lifestyle record input guidance as candidate data for improvement services. Additionally, if items requiring improvement related to weight change deviation and intake are identified together, the device (200) can generate nutrition management guidance or health checkup guidance as candidate data for improvement services.

[0453] The device (200) can store the generated improvement service candidate data by correlating it with pet identifiers, key factor data, measurement validity data, measurement deviation data, and items requiring improvement data. Through this, the device (200) can verify which measurement results and comparison results the improvement service candidate data was set on as a basis during the subsequent service matching data generation process.

[0454] Through this, the device (200) can generate improvement service candidate data by reflecting the validity of actual key factor measurements, deviations from similar populations, and items requiring improvement, rather than generating service candidates based solely on life cycle status data or care need data. Accordingly, subsequent service matching data can be generated based on candidate data that reflects the current measurement status of the pet and the need for improvement.

[0455] In step S409, the device (200) can generate lifecycle measurement management information including precision measurement condition data, measurement validity data, measurement deviation data, data on items requiring improvement, and improvement service candidate data.

[0456] Life cycle measurement management information is comprehensive management data indicating which items are measured in what manner in the current life cycle state of a pet, what meaning the measured value has when compared to statistical standards or similar populations, and what improvement service candidates it can be linked to. Life cycle measurement management information may be an intermediate processing result generated by the device (200) to reinforce the basis of service candidate data before the generation of service matching data.

[0457] The device (200) can include precision measurement condition data in lifecycle measurement management information. Since the precision measurement condition data includes the measurement cycle, measurement method, and measurement precision conditions for each key factor data, the device (200) can subsequently determine what data to collect from the guardian's terminal (100), external device, or external service server under what conditions.

[0458] The device (200) can include measurement validity data in life cycle measurement management information. The measurement validity data can indicate whether the key factor measurement data is valid, outside the standard range, requires re-measurement, or what the reliability level is when compared to statistical reference data of a group of pets belonging to the same species and life cycle.

[0459] The device (200) may include measurement deviation data in life cycle measurement management information. The measurement deviation data indicates the extent to which the key factor measurement data differs from the corresponding measurement data included in similar population data. For example, the device (200) may include the magnitude of the deviation, the direction of the deviation, whether the deviation persists, and the priority of the deviation regarding activity level, intake, bowel status, sleep amount, or weight change as measurement deviation data.

[0460] The device (200) can include data on items requiring improvement in life cycle measurement management information. The data on items requiring improvement represents items that are determined to require management or service linkage based on measurement deviation data and measurement validity data. For example, the device (200) can include insufficient activity, unbalanced intake, abnormal bowel movements, changes in sleep patterns, or deviations in weight changes as data on items requiring improvement.

[0461] The device (200) can include improvement service candidate data in life cycle measurement management information. The improvement service candidate data represents service candidates that can be used to generate subsequent service matching data in correspondence with the data of items requiring improvement. For example, the device (200) can include activity management guidance or life record input guidance as improvement service candidate data for items requiring improvement in activity level, and can include nutrition management guidance, health checkup guidance, or test kit usage guidance as improvement service candidate data for items requiring improvement in weight change.

[0462] The device (200) can store corresponding relationships between each data when generating lifecycle measurement management information. For example, the device (200) can organize a single data structure that shows how specific key factor data was measured according to certain precision measurement condition data, how the measurement value generated certain measurement validity data and measurement deviation data, and how, as a result, it was connected to certain improvement-needed item data and improvement service candidate data.

[0463] The device (200) can store life cycle measurement management information by correlating it with a pet identifier, life cycle status data, care need data, and similar population data. Through this, the device (200) can manage measurement management information for each pet separately even when the same guardian has registered multiple pets.

[0464] For example, if activity level and weight change are extracted as key factor data for a pet in its senior stage, the device (200) can generate precise measurement condition data including daily measurement conditions for activity level and weekly measurement conditions for weight. Subsequently, if the activity level measurement value is lower than the statistical standard for the same species and same life cycle group, and a continuous deviation is confirmed even when compared with similar population data, the device (200) can generate life cycle-specific measurement management information including items requiring activity level improvement and activity management guidance as improvement service candidate data.

[0465] Additionally, the device (200) can generate life cycle measurement management information that includes items requiring improvement in weight change and nutritional management guidance or health checkup guidance as candidate data for improvement services when the weight change measurement value deviates from the allowable range of statistical reference data and has a large deviation compared to weight change corresponding measurement data included in similar population data.

[0466] The device (200) can organize improvement service candidate data included in lifecycle measurement management information into a format that can be used to generate subsequent service matching data. For example, the device (200) can include service type, service identification information, improvement target items, level of provision required, basis for recommendation, and priority information in the improvement service candidate data. This information can be compared together with guardian condition data during the subsequent service matching data generation process.

[0467] The device (200) is not limited to providing life cycle measurement management information directly to the guardian's terminal (100). The life cycle measurement management information may be used as internal processing data used to generate service matching data, and may be provided to the guardian's terminal (100) in the form of a measurement request, a measurement result notification, or a service candidate notification, depending on the embodiment.

[0468] Through this, the device (200) can integrate the selection of key factors, the setting of precise measurement conditions, the determination of the validity of actual measurement values, the analysis of deviations compared to similar populations, and the generation of improvement service candidates into a single life cycle measurement management information. Accordingly, subsequent service matching data is not generated based solely on simple life cycle status or care need values, but can be generated based on data reflecting the actual measurement results of the pet and the need for improvement.

[0469] Consequently, the device (200) can improve the basis and accuracy of service matching data by selecting high-impact measurement items in the life cycle status and care needs of pets, and generating improvement service candidate data by reflecting statistical criteria and comparison results with similar populations.

[0470] FIG. 5 is a flowchart illustrating the process of generating end-of-life care information according to one embodiment.

[0471] Specifically, after generating service matching data, the device (200) can generate end-of-life management information to organize the service matching data into a form that can actually provide services related to end-of-life management.

[0472] End-of-life care information is data that indicates the availability of end-of-life care services, execution priority, scheduled execution time, guardian input request conditions, and guardian notification conditions by synthesizing the pet's life cycle status, level of end-of-life care need, pet characteristics, health limitations, guardian detailed conditions, and provision conditions of target services.

[0473] After generating service matching data, the device (200) can identify target service data corresponding to the end-of-life care service type among the services included in the service matching data. The end-of-life care service type may include at least one of a pet funeral preparation type, a pet transportation support type, a pet post-mortem care type, a memorial content creation type, and a memorial space usage type.

[0474] After generating end-of-life management information, the device (200) may include the end-of-life management information in the service provision information. In this case, the service provision information may not be a simple service list, but may be composed of information including end-of-life management stages, available services, execution priority, scheduled execution time, guardian input request conditions, and notification conditions that can be provided to the guardian's terminal (100).

[0475] For example, if the device (200) has a high level of need for end-of-life care, confirms information on pet movement restrictions, and includes specific burial method conditions and transport conditions in the guardian condition data, it can identify transport support services, burial preparation services, and memorial content creation services among end-of-life care services as target services. Subsequently, the device (200) generates end-of-life care information by reflecting the service availability area, preparation period, information required for guardian input, and execution priority for each service, and can include this in the service provision information provided to the guardian's terminal (100).

[0476] Through this, the device (200) can ensure that the end-of-life care service is not merely displayed as a recommendation list, but is organized into actionable information tailored to the condition of the pet and the guardian. Accordingly, the service provision information provided to the guardian's terminal (100) can be composed of specific guidance information that includes both the end-of-life care stage and the service execution conditions.

[0477] Referring to FIG. 5, first, in step S501, the device (200) can extract the level of end-of-life care need from the care need data.

[0478] The level of need for end-of-life care is data indicating the degree of need to provide end-of-life care services for a companion animal. The level of need for end-of-life care may consist of at least one of the end-of-life care need value, end-of-life related care grade, observation need status value, or whether there is a need for linkage with end-of-life care services, which are included in the care need data.

[0479] The device (200) can identify items related to end-of-life care among the previously calculated care needs data. For example, if the care needs data includes at least one of health care needs, nutrition care needs, activity care needs, medical examination care needs, end-of-life care needs, and aftercare needs, the device (200) can extract a value corresponding to the end-of-life care needs among these as the end-of-life care needs level.

[0480] The device (200) can extract the level of need for end-of-life care as a numerical value, a grade value, or a status value. For example, if the value of the need for end-of-life care is calculated as a score in the range of 0 to 100, the device (200) can extract that score as the level of need for end-of-life care. Additionally, if the need for end-of-life care is calculated as a grade value such as low, normal, high, or need for urgent verification, the device (200) can extract that grade value as the level of need for end-of-life care.

[0481] The device (200) can check the correspondence with life cycle status data when extracting the level of need for end-of-life care. For example, if the life cycle status data corresponds to old age, a state requiring observation, or a state requiring end-of-life care, and the value of the need for end-of-life care in the care need data is greater than or equal to a preset standard, the device (200) can extract the level of need for end-of-life care as a high level.

[0482] The device (200) is not limited to a configuration that simply extracts the level of end-of-life care required as a single numerical value. The device (200) can extract information on care items that serve as the basis for calculating the level of end-of-life care required, along with the level of end-of-life care required. For example, the device (200) can identify which items among reduced activity levels, weight changes, disease status, potential for mobility limitations, changes in lifestyle patterns, or caregiver observation records influenced the calculation of the level of end-of-life care required.

[0483] The device (200) can extract the level of end-of-life care required and store it by correlating it with a pet identifier, life cycle status data, and care need data. Through this, the device (200) can reliably reference the level of end-of-life care required for the same pet when generating end-of-life care stage data thereafter.

[0484] For example, the device (200) can extract the end-of-life care need level if, for a specific pet, the end-of-life care need value in the care need data is calculated as 82 points and the life cycle status data corresponds to the old age or end-of-life care need status. These numbers and grades are examples for illustrative purposes, and the device (200) may use different score ranges or grade systems depending on the embodiment.

[0485] Through this, the device (200) can obtain quantitative or graded data on how much end-of-life care service is needed for a pet before starting to generate end-of-life care information. Accordingly, in subsequent processing, it can prepare basic data for determining the end-of-life care stage and execution priority, along with pet characteristic data, health limitation data, and service provision conditions.

[0486] In step S502, the device (200) can extract pet characteristic data including species, breed, weight, and age from pet registration data.

[0487] Pet characteristic data is data used to determine the basic conditions of a target pet during the process of generating end-of-life care information. Pet characteristic data includes species, breed, weight, and age, and may further include gender, neutering status, body shape information, or a pet identifier depending on the embodiment.

[0488] The device (200) can extract species and breeds from registered pet data through the guardian's terminal (100). The species and breeds can be used when applying criteria for determining the end-of-life care stage or when comparing the conditions of the pets eligible for the end-of-life care service. For example, the device (200) can determine whether the pet is a dog or a cat, or whether it belongs to a specific breed such as a Maltese, Poodle, or Korean Shorthair.

[0489] The device (200) can extract weight from pet registration data. The weight can be used to determine the conditions of the target pet related to the pet transport support type, the pet funeral preparation type, or the pet post-mortem care type. If the weight is entered in a different unit, the device (200) can convert it to a weight value in the same unit according to a preset unit conversion standard.

[0490] The device (200) can extract age from pet registration data. The age may be a value directly entered through the guardian's terminal (100) or a value calculated based on the date of birth. The device (200) can use the age together with life cycle status data to reflect it in the generation of end-of-life management stage data.

[0491] For example, the device (200) can generate pet characteristic data by extracting information from pet registration data such as the species being dog, the breed being Maltese, the weight being 4.2 kg, and the age being 12 years. The device (200) can compare the generated pet characteristic data with the end-of-life stage determination criteria and target pet conditions in subsequent processing.

[0492] The device (200) can check for missing pet characteristic data or format errors. For example, if weight information is missing or determined to be a value entered a long time ago, the device (200) can provide a request to the guardian's terminal (100) to input the latest weight. Additionally, if breed information is unclear, the device (200) can set an alternative classification value based on the breed information and weight information.

[0493] Through this, the device (200) can obtain basic characteristics of the pet necessary for generating end-of-life care information as structured data. Accordingly, the device (200) can generate end-of-life care stage data, compare conditions of target service data, and calculate pet suitability data based on objective characteristic values ​​such as species, breed, weight, and age.

[0494] In step S503, the device (200) can extract health restriction data including disease status information and movement restriction information from health history data and life record data.

[0495] Health limitation data is data used to determine health limitations of a pet during the process of generating end-of-life care information. Health limitation data includes information on disease status and mobility limitations, and, depending on the embodiment, may further include records related to pain, records of reduced activity, records of decreased appetite, records related to respiration, records of abnormal bowel movements, or observation notes from the guardian.

[0496] Disease status information is data indicating the current or past disease status of a pet. The device (200) may extract at least one of a diagnosis name, disease type, degree of disease progression, whether it is a chronic disease, recent medical results, test results, medication history, or surgery history from health history data and use it as disease status information.

[0497] Movement restriction information is data indicating a restriction state that must be considered during the process of moving or transporting a pet. The device (200) may extract at least one of the following from the living record data and use it as movement restriction information: whether there is difficulty walking, whether there is a sharp decrease in activity level, whether there is a decrease in walking time, whether there is difficulty moving outside, or movement-related observation records entered through the guardian's terminal (100).

[0498] The device (200) can generate health limitation data by using health history data and lifestyle record data together. For example, if a history of joint disease is identified in the health history data and a record of reduced walking time or difficulty moving up stairs is identified in the lifestyle record data, the device (200) can construct health limitation data by correlating the disease status information and the movement limitation information with each other.

[0499] The device (200) is not limited to managing disease status information and movement restriction information only as simple text records. The device (200) can convert disease status information and movement restriction information into numerical values, grade values, or status values ​​according to preset health restriction classification criteria. For example, disease status information can be converted into grade values ​​such as mild, need observation, need continuous management, or need urgent verification, and movement restriction information can be converted into status values ​​such as none, mild, moderate, severe, or need movement assistance.

[0500] The device (200) can verify the recency and source of the data when extracting health restriction data. For example, if only a disease history recorded a long time ago exists and a restriction status is not confirmed in recent lifestyle records, the device (200) can classify the relevant disease status information as past history. Conversely, if the same restriction status is repeatedly confirmed in recent medical results and lifestyle records, the device (200) can set a high priority for reflecting the relevant health restriction data.

[0501] For example, if the device (200) identifies a history of heart disease and a history of recent medication in the health history data, and identifies a decrease in activity level and a decrease in walking time in the lifestyle record data, it can generate health restriction data including disease status information and movement restriction information. This health restriction data can subsequently be used to generate end-of-life care stage data, compare the conditions of target pets, and calculate pet suitability data.

[0502] Through this, the device (200) can reflect the species, breed, weight, and age of the pet, as well as disease conditions and mobility restriction conditions that may affect the actual method of service provision. Accordingly, the feasibility, method of provision, and priority of execution of end-of-life care services can be determined based on more realistic conditions.

[0503] In step S504, the device (200) can generate end-of-life care stage data by comparing life cycle status data, end-of-life care need level, pet characteristic data, and health limitation data with preset end-of-life stage determination criteria.

[0504] The criteria for determining the end-of-life stage are pre-established standards for determining which management stage a companion animal's current condition corresponds to in the end-of-life management process. The criteria for determining the end-of-life stage may include at least one of the following: criteria for life cycle status, criteria for the level of need for end-of-life management, age criteria by species or breed, weight criteria, disease status criteria, and movement restriction criteria.

[0505] End-of-life care stage data is data indicating the need for preparation or implementation of end-of-life care services. End-of-life care stage data can be generated as stage values, grade values, score values, or status values ​​corresponding to at least one of the observation need stage, pre-preparation stage, intensive care stage, emergency preparation stage, or post-care linkage preparation stage.

[0506] The device (200) can compare life cycle status data and the level of need for end-of-life care with the end-of-life stage determination criteria. For example, if the life cycle status data corresponds to old age, a state requiring observation, or a state requiring end-of-life care, and the level of need for end-of-life care is greater than or equal to a preset standard, the device (200) can set a candidate stage corresponding to a preparatory stage or an intensive care stage.

[0507] The device (200) can correct candidate stages by reflecting pet characteristic data and health limitation data together. For example, if elderly or a specific weight range is identified in the pet characteristic data, and severe disease status information, recurring medical care information, difficulty walking information, or a need for mobility assistance is identified in the health limitation data, the device (200) can generate end-of-life care stage data at a higher care stage.

[0508] The device (200) is not limited to a configuration that generates end-of-life care stage data based on only one criterion. The device (200) can calculate judgment scores corresponding to life cycle status data, end-of-life care need level, pet characteristic data, and health limitation data, respectively, and generate end-of-life care stage data by combining the calculated judgment scores.

[0509] For example, the device (200) can generate end-of-life care stage data corresponding to a preparatory stage or an intensive care stage when the life cycle status data corresponds to the elderly stage, the level of need for end-of-life care is high, and mobility restriction information is confirmed in the health restriction data. Additionally, when both disease status information and mobility restriction information are confirmed to be at a high level, it can generate end-of-life care stage data corresponding to an emergency preparation stage.

[0510] The device (200) is not limited to generating end-of-life care stage data as a medical diagnosis result. The end-of-life care stage data may not be a medical judgment for diagnosing or treating a pet's disease, but rather platform processing data for setting the availability of end-of-life care services, the need for preparation, the priority of execution, and the guardian notification conditions.

[0511] The device (200) can store the generated end-of-life care stage data by correlating it with the end-of-life care need level, pet characteristic data, and health restriction data. Accordingly, the device (200) can refer to the end-of-life care stage data during the subsequent processes of extracting target service data, selecting available service data, calculating execution priorities, and setting scheduled execution times for each service.

[0512] Through this, the device (200) can generate end-of-life care stage data by considering the pet's life cycle status, basic characteristics, and health limitations together, rather than simply recommending services based only on the level of need for end-of-life care. Accordingly, subsequent end-of-life care information can be composed of data that reflects the pet's actual condition and the feasibility of service execution.

[0513] In step S505, the device (200) can extract target service data corresponding to the end-of-life care service type from the service matching data.

[0514] Target service data is data representing a service among the services included in the service matching data that is the target for generating end-of-life management information. Target service data may include at least one of service identification information, service type, service name, availability of provision, provision priority candidate value, and service provision condition identification information.

[0515] The device (200) can check service type information included in service matching data and extract a service corresponding to an end-of-life care service type among them as target service data. The end-of-life care service type may include at least one of a pet funeral preparation type, a pet transportation support type, a pet post-mortem care type, a memorial content creation type, and a memorial space usage type.

[0516] In this case, the type of post-mortem care for the pet may refer to the type of support for post-mortem procedures necessary for preparing for burial, transportation, guardian verification procedures, or linking memorial services after the death of the pet.

[0517] For example, if the service matching data includes health checkup guidance, nutrition management guidance, funeral preparation guidance, transportation support guidance, and memorial content creation guidance, the device (200) can extract the funeral preparation guidance, transportation support guidance, and memorial content creation guidance among these as target service data corresponding to the end-of-life care service type.

[0518] Even if the service type is not explicitly stored, the device (200) can determine whether it corresponds to an end-of-life care service type based on service identification information, service name, service classification code, or service provision condition identification information. For example, if the service classification code corresponds to funeral preparation or the use of a memorial space, the device (200) can extract the corresponding service as target service data.

[0519] When the device (200) extracts target service data, it may extract all end-of-life care service candidates included in the service matching data, or it may extract only some services that satisfy preset extraction criteria. The extraction criteria may include at least one of a service matching score, availability of service provision, relevance to end-of-life care stage data, service type priority, or basic satisfaction with guardian condition data.

[0520] After extracting target service data, the device (200) can correlate the target service data with service provision condition identification information so that provision condition data can be verified in subsequent processing. Accordingly, the device (200) can subsequently extract target pet conditions, preparation time, available provision areas, provision method conditions, and guardian input requirements for each target service data.

[0521] Through this, the device (200) can select only the services corresponding to the end-of-life management service type and set them as targets for subsequent judgment, rather than using the entire service matching data collectively for generating end-of-life management information. Accordingly, end-of-life management information can be generated more accurately for services related to funeral preparation, transportation support, post-mortem care, creation of memorial content, or use of memorial spaces.

[0522] In step S506, the device (200) can extract provision condition data from the target service data, including target pet conditions, preparation time, available regions, provision method conditions, and guardian input requirements.

[0523] The provision condition data is data indicating the conditions under which the end-of-life care service included in the target service data can actually be provided. The provision condition data may include conditions for the target pet, the preparation period, available service areas, service method conditions, and information required for guardian input, and may further include cost conditions, available reservation times, service provider information, or provision restriction conditions depending on the embodiment.

[0524] Eligible pet conditions refer to the conditions of a pet for which specific end-of-life care services may be provided. Eligible pet conditions may include at least one of species, breed, weight range, age range, health status, mobility, or end-of-life care stage. For example, services of the pet transport support type may include weight range, mobility restriction status, or transportable area as eligible pet conditions.

[0525] The preparation period is data representing the preparation time required before the target service is actually executed. For example, the pet funeral preparation type may include the time required for reservation confirmation, verification of guardian input information, or site assignment as the preparation period, while the memorial content creation type may include the time required to collect and process photos, names, messages, or content creation materials as the preparation period.

[0526] The available area is data indicating the geographical range where the target service can be provided. The available area may be set as at least one of a city / province, a city / county / district, a service zone, a transportable distance, or an area where a specific facility is available. The device (200) can compare the available area with the geographical conditions included in the guardian detailed condition data.

[0527] The provision method condition is a condition regarding the method in which the target service is provided. The provision method condition may include at least one of a visit provision method, a guardian visit method, a transportation support method, an online provision method, an offline facility usage method, or a content provision method. For example, the memorial content creation type may be set as an online content provision method, and the memorial space usage type may be set as an offline facility usage method.

[0528] Information required for guardian input refers to information that must be additionally received from the guardian in order to provide the target service. Information required for guardian input may include at least one of the following: guardian contact information, desired schedule, information on selection of funeral method, information on transportation location, photos of the pet, memorial message, information on selection of memorial content format, or information on consent to use the service.

[0529] The device (200) can extract provision condition data using service identification information or service provision condition identification information included in the target service data. For example, if the target service data corresponds to a pet funeral preparation type, the device (200) can query the target pet conditions, preparation period, available service area, provision method conditions, and guardian input requirements corresponding to the service from the service condition database.

[0530] For example, if the target service data is a transport support service, the device (200) can extract weight range and movement restriction status conditions as target pet conditions, extract transportable zones as available areas, and extract departure location, desired transport time, and guardian contact information as guardian input required information. If the target service data is a memorial content creation service, the device (200) can extract preparation time, content provision method conditions, and information required to input pet photos and memorial messages as provision condition data.

[0531] The device (200) can store the extracted provision condition data by correlating it with the target service data. Accordingly, in subsequent processing, the device (200) can compare the pet characteristic data and health restriction data with the target pet conditions, and compare the guardian detailed condition data with the provision available region, provision method conditions, and guardian input required information.

[0532] Through this, the device (200) can determine whether end-of-life care services can be provided not merely based on the type of service, but specifically based on the conditions of the target pet, the preparation period, the areas where they can be provided, the conditions of the method of provision, and the information required for guardian input. Accordingly, in subsequent steps, the data on available services, the priority of execution, and the scheduled time of execution for each service can be calculated more accurately.

[0533] In step S507, the device (200) can extract guardian detailed condition data including regional conditions, pet burial method conditions, pet transport conditions, and memorial content conditions from guardian condition data.

[0534] Guardian detailed condition data is detailed condition data extracted from guardian condition data to determine the availability and method of providing end-of-life care services. Guardian detailed condition data may include regional conditions, pet burial method conditions, pet transportation conditions, and memorial content conditions, and may further include budget conditions, desired schedule conditions, notification reception conditions, or service provision method preference conditions depending on the embodiment.

[0535] The regional condition is data indicating the geographical area where a guardian can access end-of-life care services. The regional condition may include at least one of the guardian's address, the desired service area, the pet's current location, the transfer origin, the transfer destination, or the area of ​​available facilities. The device (200) can be used to compare the regional condition with the available areas included in the provision condition data.

[0536] The pet burial method condition is data regarding the burial method preferred or selected by the guardian. For example, the pet burial method condition may include at least one of cremation, interment, tree burial, use of a memorial space, or use of a burial facility. The device (200) can compare this condition with target service data for the pet burial preparation type or the memorial space use type.

[0537] The pet transport condition is a condition requested by the guardian regarding the movement or transport support of the pet. The pet transport condition may include at least one of whether pickup is required, the desired time of transport, the departure location, whether the guardian accompanies the pet, the travel distance condition, or the method of transport support. The device (200) can compare the transport condition with the conditions for the method of providing the type of pet transport support and the available area.

[0538] The memorial content conditions are conditions set by the guardian in relation to the creation or provision of memorial content. The memorial content conditions may include at least one of whether to use a photo of the pet, whether to enter a memorial message, whether to use a mobile memorial space, whether to use a guestbook, the format of the memorial content, the scope of public disclosure of the memorial content, or whether to receive a memorial notification.

[0539] The device (200) can extract guardian detailed condition data from guardian condition data that is input or modified through the guardian's terminal (100). For example, if the guardian inputs an available area, a burial method, whether a pickup service is required, and whether memorial content is received on the guardian's terminal (100), the device (200) can extract the input values ​​as a region condition, a pet burial method condition, a pet transport condition, and a memorial content condition, respectively.

[0540] The device (200) can check for missing guardian detailed condition data or format errors. For example, if a departure location is required to use the transport support service but such information is not entered, the device (200) can provide a request for additional input to the guardian's terminal (100). Additionally, if a photo or memorial phrase required for creating memorial content is missing, the item can be marked as an item requiring guardian input.

[0541] The device (200) can store the extracted guardian detailed condition data by correlating it with guardian identification information, pet identifier, and target service data. Accordingly, the device (200) can then calculate condition suitability data by comparing the provided condition data with the guardian detailed condition data.

[0542] Through this, the device (200) can determine whether end-of-life care services are available based solely on conditions on the service provider's side, but can also reflect the guardian's region, burial method, transportation request, and preference for memorial content. Accordingly, in subsequent processing, it can more accurately select end-of-life care services that are actually available to the guardian.

[0543] In step S508, the device (200) can calculate pet suitability data by comparing pet characteristic data and health limitation data with the target pet condition.

[0544] Pet suitability data is data indicating whether target service data can be provided to a specific pet, or whether there are restrictions on provision. Pet suitability data may include at least one of suitability, a suitability score, a reason for restriction, a grade of availability, or whether additional verification is required.

[0545] The target pet condition is the condition of a pet to which the end-of-life care service included in the target service data can be applied. The target pet condition may include at least one of species, breed, weight range, age range, condition allowing disease status, condition allowing movement restriction, or end-of-life care stage condition.

[0546] The device (200) can compare the species, breed, weight, and age included in the pet characteristic data with the target pet conditions. For example, if a specific transport support service is set for pets with a weight below a certain threshold, the device (200) can check whether the weight included in the pet characteristic data falls within the corresponding weight range.

[0547] The device (200) can compare disease status information and movement restriction information included in health restriction data with the target pet conditions. For example, if the target service data is a visiting transport support service and the target pet conditions include a movement restriction state allowed or a movement support required state, the device (200) can determine whether the movement restriction information included in the health restriction data satisfies the corresponding conditions.

[0548] The device (200) can calculate pet suitability data by comparing pet characteristic data and health restriction data with individual conditions, respectively, and then combining the comparison results. For example, if all conditions regarding species, weight, and movement restriction are satisfied, the device (200) can calculate a suitable state or a high suitability score for the corresponding target service data. Conversely, if the weight range is exceeded or a change in the provision method is required due to a specific disease condition, the device (200) can calculate pet suitability data including the reason for restriction or a state requiring additional verification.

[0549] The device (200) is not limited to a configuration that calculates pet suitability data as simple binary values ​​of possible or impossible. The device (200) can calculate a suitability score by distinguishing between whether essential conditions are met and the degree to which optional conditions are met. For example, species and weight conditions may be determined as essential conditions, and breed or age conditions may be reflected as correction items for the suitability score.

[0550] For example, if the pet characteristic data is a small dog, weight 4.2 kg, and age 12 years, and the health limitation data includes a condition requiring walking difficulty and mobility support, the device (200) can produce high pet suitability data compared to the target pet conditions of the transport support service. On the other hand, if the weight condition of the target service data is exceeded, it can produce pet suitability data indicating a low probability of provision or the need for separate verification.

[0551] The device (200) can correlate the calculated pet suitability data with the target service data. Accordingly, the device (200) can then compare the guardian detailed condition data with the provision condition data to calculate the condition suitability data, and use the pet suitability data and the condition suitability data together to select the service data that can be provided.

[0552] Through this, the device (200) can first verify the possibility of providing end-of-life care services based on the pet's species, breed, weight, age, disease status, and mobility restriction status, rather than judging the possibility based solely on the guardian's selection criteria. Accordingly, end-of-life care services applicable to the actual pet can be selected more accurately in a subsequent step.

[0553] In step S509, the device (200) can calculate condition suitability data by comparing guardian detailed condition data with provided condition data.

[0554] Condition suitability data is data indicating the extent to which the provision conditions of the target service data match the guardian's usage conditions. Condition suitability data may include at least one of suitability, suitability score, items that meet conditions, items that do not meet conditions, whether supplementary input is required, or whether an alternative service is required.

[0555] The device (200) can compare the regional conditions included in the guardian detailed condition data with the available regions included in the provision condition data. For example, if the regional conditions entered by the guardian are included in the available regions of the target service, the device (200) can determine that the regional conditions are satisfied. Conversely, if they fall outside the available regions, the device (200) can include a state of regional non-satisfaction or the need to check for an alternative region in the condition suitability data.

[0556] The device (200) can compare the pet burial method conditions included in the guardian detailed condition data with the provision method conditions of the provision condition data. For example, if the guardian has selected a specific burial method or the use of a memorial space, and the target service data supports the provision conditions of that method, the device (200) can determine that the burial method conditions are satisfied.

[0557] The device (200) can compare the pet transport conditions with the available regions, conditions of the method of provision, and the preparation period of the provision condition data. For example, if a guardian requests a pickup, the device (200) can check whether the target service provides a transport support method, whether the departure location entered by the guardian is included in the available regions, and whether the desired transport time is possible considering the preparation period.

[0558] The device (200) can compare the memorial content conditions with the provision method conditions of the provision condition data and the guardian input requirements information. For example, if the guardian has allowed the creation of a mobile memorial space or memorial content, the device (200) can check whether the target service data supports the content provision method and whether guardian input requirements information, such as pet photos, memorial phrases, or privacy settings, are satisfied.

[0559] The device (200) can calculate condition suitability data by combining the comparison results for each condition. For example, if the regional condition, burial method condition, transportation condition, and memorial content condition are all satisfied, the device (200) can calculate a high condition suitability score for the corresponding target service data. If some conditions are not satisfied, the device can calculate condition suitability data including unsatisfied items and items requiring additional input.

[0560] The device (200) is not limited to a configuration that calculates condition suitability data as simple binary values ​​of possible or impossible. The device (200) can calculate a suitability score by distinguishing between mandatory conditions and optional conditions. For example, available regions and mandatory input information can be reflected as mandatory conditions, while the format of memorial content or notification method can be reflected as optional conditions.

[0561] For example, if the guardian's regional condition is included in the serviceable area of ​​the target service, the funeral method selected by the guardian matches the serviceable method condition of the target service, and the transfer departure location and desired schedule satisfy the service condition, the device (200) can produce high condition suitability data for the target service data. Conversely, if photos or memorial phrases required for creating memorial content are not entered, the device (200) can include a guardian input requirement state in the condition suitability data.

[0562] The device (200) can correlate the calculated condition suitability data with the target service data and the guardian detailed condition data. Accordingly, the device (200) can subsequently select the service data that can be provided among the target service data by using the pet suitability data and the condition suitability data together.

[0563] Through this, the device (200) can determine the feasibility of providing end-of-life care services not solely based on conditions on the service provider's side, but by comparing it with the guardian's region, burial method, transportation request, and memorial content conditions. Accordingly, in a subsequent step, it can more accurately select end-of-life care services that are actually available to the guardian.

[0564] In step S510, the device (200) can select available service data among the target service data based on pet suitability data and condition suitability data.

[0565] Available service data is data representing services that can be set as actual targets for provision among the target service data, considering the current characteristics of the pet, health restriction status, and detailed conditions of the guardian. Available service data may include at least one of service identification information, service type, availability, reason for restriction, candidate values ​​for provision priority, and whether additional input is required.

[0566] The device (200) can check the pet suitability data and the condition suitability data together. For example, if the target service data satisfies the conditions for the pet's species, weight, age, disease status, and movement restriction, and simultaneously satisfies the guardian's regional conditions, burial method conditions, transportation conditions, and memorial content conditions, the device (200) can select the target service data as available service data.

[0567] The device (200) is not limited to a configuration that selects available service data based on only one of the pet suitability data and condition suitability data. For example, even if the pet suitability data is high, if the guardian's regional condition falls outside the available service area, the service may be excluded from available service data or classified as a target for alternative service verification. Conversely, even if the guardian condition is met, if the pet's weight, disease status, or movement restriction status does not satisfy the target pet condition, the service may be excluded from available service data.

[0568] The device (200) may apply pre-set selection criteria to pet suitability data and condition suitability data. The selection criteria may include at least one of a pet suitability score criterion, a condition suitability score criterion, whether essential conditions are met, whether a reason for restriction exists, whether additional input is required, and whether an alternative service exists.

[0569] For example, the device (200) can select target service data as available service data in which both the pet suitability score and the condition suitability score are above a preset standard. Additionally, if the essential conditions are met but some of the information required for guardian input is missing, the device (200) may maintain the service as available service data while also assigning a status requiring additional input.

[0570] When multiple target service data are selected as service data that can be provided, the device (200) may store the reasons for the availability and the reasons for the restriction for each service. For example, a funeral preparation service may be selected as service data that can be provided by satisfying regional conditions and funeral method conditions, and a memorial content creation service may be selected as service data that can be provided with the condition that a photo or memorial text input is required.

[0571] The device (200) can correspond the available service data with a pet identifier, end-of-life care stage data, pet suitability data, and condition suitability data. Accordingly, the device (200) can perform subsequent processing only on the services that are actually available during the process of calculating execution priorities and setting scheduled execution times for each service.

[0572] Through this, the device (200) does not determine the target of the service solely on the grounds that it corresponds to the end-of-life care service type, but can select service data that can be provided by verifying together the applicability of the pet and whether the guardian conditions are met. Accordingly, the end-of-life care information provided to the guardian's terminal (100) can be configured around services that are highly likely to be actually used.

[0573] In step S511, the device (200) can calculate the execution priority of available service data based on end-of-life care stage data and end-of-life care need level.

[0574] Execution priority is data indicating the order or importance with which each service included in the available service data should be executed. Execution priority may consist of at least one of a priority grade, a priority score, an execution sequence number, whether it is an urgent process, or whether it is a preceding service.

[0575] The device (200) can set the basic priority of service data available for provision based on end-of-life care stage data. For example, if the end-of-life care stage data corresponds to the pre-preparation stage, the device (200) can set services related to selecting a burial method, verifying guardian input information, or preparing memorial content as priority processing targets. If the end-of-life care stage data corresponds to the intensive care stage or the emergency preparation stage, the device (200) can set services related to transportation support, burial preparation, or guardian notification as a relatively high priority.

[0576] The device (200) can adjust the degree of priority using the level of end-of-life care need. For example, if the level of end-of-life care need is high, the device (200) can set services that require a long preparation period or guardian input among the available service data as priority targets for execution. Conversely, if the level of end-of-life care need is low or the level of observation need is low, the device (200) can prioritize the placement of guidance-type services or pre-preparation-type services.

[0577] The device (200) can calculate execution priorities by reflecting the order of service types. For example, the pet funeral preparation type may require securing necessary information before the pet post-mortem care type or the memorial space usage type, and the memorial content creation type may be executed only after pet photos, memorial phrases, or privacy settings are entered first. In this case, the device (200) can set the service requiring prior input or prior reservation to a high priority.

[0578] The device (200) can calculate a priority score for each available service data. The priority score can be calculated by combining at least one of a stage score corresponding to end-of-life care stage data, a need score corresponding to the level of need for end-of-life care, an urgency score by service type, a correction score based on the preparation period, and a correction score based on whether guardian input is required.

[0579] For example, if the end-of-life care stage data is an emergency preparation stage and the level of need for end-of-life care is high, the device (200) can calculate a high priority for the execution of the transport support service and the funeral preparation service. Additionally, even if the memorial content creation service is included in the available service data, if the execution time can be relatively late, it can be set to a lower priority than the funeral preparation service.

[0580] The device (200) may additionally apply alignment criteria among services assigned the same priority when calculating execution priority. The alignment criteria may include at least one of the preparation time, the number of guardian input required information, the degree of restriction on available areas, the reservation time, or the service type priority.

[0581] The device (200) can store the calculated execution priority in correspondence with the available service data. Accordingly, when setting the scheduled execution time for each service thereafter, the device (200) can reflect the preparation period and guardian input request conditions starting from the service with the highest priority.

[0582] Through this, the device (200) can determine the execution order of each service according to the end-of-life care stage and the level of care required of the pet, rather than simply listing the available service data. Accordingly, the end-of-life care information can be composed of actionable information that reflects urgency, the need for preparation, and the order of precedence between services.

[0583] In step S512, the device (200) can set the scheduled execution time for each service based on the execution priority and the preparation period.

[0584] The scheduled execution time per service refers to the scheduled time or scheduled period during which each service included in the available service data is set to be notified to, scheduled for, or executed by the guardian. The scheduled execution time per service may include at least one of the following: a scheduled execution date, a scheduled execution time zone, a recommended reservation time, a guardian input deadline, or a notification provision time.

[0585] The device (200) can set the scheduled time for execution by reflecting the preparation period starting from the service with the highest execution priority. For example, if the execution priority of the funeral preparation service is high and the preparation period is set to 3 days, the device (200) can set the scheduled time for execution of the funeral preparation service by considering the guardian input information, the availability of reservation, and the service execution date.

[0586] The device (200) may prioritize the placement of services that require a long preparation period. For example, if a service of the memorial content creation type requires pet photos, memorial phrases, or content format selection information and requires a certain production period, the device (200) may set the time of the guardian input request for the service ahead of the scheduled time of actual content provision.

[0587] The device (200) can adjust the scheduled execution time for each service by considering the order of precedence between services. For example, since the pet transport support type can be linked to the funeral preparation type, the device (200) can set the scheduled transport time so that it is not later than the scheduled execution time of the funeral preparation service. Additionally, the memorial space usage type or the memorial content creation type can be placed at a time after the funeral preparation or post-mortem care related services.

[0588] If the end-of-life care stage data corresponds to the emergency preparation stage, the device (200) can set the scheduled execution time of a service with a high execution priority to an earlier time. Conversely, if the end-of-life care stage data corresponds to the preliminary preparation stage, the device (200) can set the scheduled execution time in stages by considering the guardian input and the reservation availability period.

[0589] For example, if the available service data includes a transport support service, a funeral preparation service, and a memorial content creation service, the device (200) may first set the scheduled execution time for the transport support service and the funeral preparation service according to the end-of-life care stage data and execution priority, and then set the scheduled execution time for the memorial content creation service to a later schedule.

[0590] The device (200) can store the scheduled execution time for each set service by correlating it with available service data, execution priority, and preparation time. Accordingly, when the device (200) sets the guardian input request condition and guardian notification condition in subsequent processing, it can calculate the input request time and the notification time in accordance with the scheduled execution time of each service.

[0591] Through this, the device (200) can specifically set the scheduled time for execution of each service by reflecting the execution priority and the preparation period, rather than merely providing guidance on whether end-of-life care services are available. Accordingly, end-of-life care information can be composed of schedule-based information that allows the caregiver to perform necessary input and preparation at the necessary time.

[0592] In step S513, the device (200) can set guardian input request conditions and guardian notification conditions based on guardian input required information and service-specific execution times.

[0593] The guardian input request condition is a condition for requesting additional input from the guardian's terminal (100) to execute the end-of-life management service. The guardian input request condition may include at least one of an input request target item, an input request time, an input deadline time, a re-request condition, whether input is required, and a processing condition for missing input.

[0594] The guardian notification condition is a condition for providing a notification related to end-of-life care information to the guardian's terminal (100). The guardian notification condition may include at least one of the following: a notification provision time, a notification period, a notification method, a notification target service, a notification content, a re-notification condition, and a notification suspension condition.

[0595] The device (200) can identify the guardian input requirements included in the provision condition data and distinguish the input items required for each target service data or available service data. For example, the pet funeral preparation type may require information on selecting the funeral method, desired schedule, guardian contact information, or consent to use the facility, and the pet transportation support type may require a departure location, desired transportation time, and whether the guardian accompanies the pet.

[0596] The device (200) can set the time for requesting guardian input and the time for input deadline based on the scheduled time for execution per service. For example, if the memorial content creation type requires a pet photo and a memorial message before the scheduled time for execution, the device (200) can set the guardian input request condition at a time earlier than the scheduled time for execution, taking into account the time required for content creation.

[0597] The device (200) can set different request conditions depending on the importance of the information required for guardian input. For example, contact information, location information, or funeral method selection information that is essential for providing the service can be set as required input items, and memorial phrases, photo selection, or disclosure scope settings can be set as optional input items depending on the service type.

[0598] The device (200) can set conditions for a re-request when guardian input information has not been entered. For example, if a certain amount of time remains until the scheduled execution time and required input items are missing, the device (200) can set conditions for a guardian input request to provide a re-entry request notification to the guardian's terminal (100). Additionally, if the input deadline has passed, the service can be processed in a pending state or a state requiring additional verification.

[0599] The device (200) can set guardian notification conditions by reflecting the scheduled execution time and execution priority for each service. For example, for a transportation support service or a funeral preparation service with a high execution priority, it can be set so that multiple notifications are provided before the scheduled execution time, and for a memorial content creation service, it can be set so that a request for submission of input materials or a notification of scheduled content provision is provided.

[0600] For example, if the available service data includes a transport support service and a memorial content creation service, the device (200) may set conditions for requesting input of a departure location and a desired transport time for the transport support service, and conditions for requesting input of a pet photo and a memorial message for the memorial content creation service. Additionally, the device (200) may set guardian notification conditions so that a pre-notification, an input deadline notification, or an execution schedule notification is provided to the guardian's terminal (100) in accordance with the scheduled execution time for each service.

[0601] The device (200) can store service data capable of providing set guardian input request conditions and guardian notification conditions, scheduled execution times for each service, and information requiring guardian input. Accordingly, when generating end-of-life management information, the device (200) can include the guardian input items and notification provision methods required for each service.

[0602] Through this, the device (200) can not only simply provide guidance on end-of-life care services but can also specifically set the timing for guardian input and notification provision required before the service is executed. Accordingly, the guardian can input necessary information at an appropriate time through the guardian's terminal (100), and the device (200) can support the end-of-life care service to be executed at the scheduled time.

[0603] In step S514, the device (200) can generate end-of-life management information including end-of-life management stage data, available service data, execution priority, scheduled execution time per service, guardian input request conditions, and guardian notification conditions.

[0604] End-of-life management information is comprehensive data provided to a guardian's terminal (100) in relation to the end-of-life management of a pet, or used for service execution management of a device (200). End-of-life management information may include the pet's current management stage, actual services available, execution order by service, scheduled execution time, guardian input request items, and notification conditions.

[0605] The device (200) can include end-of-life care stage data in end-of-life care information. The end-of-life care stage data may indicate the level of end-of-life care currently required for the pet, such as the observation required stage, the pre-preparation stage, the intensive care stage, the emergency preparation stage, or the post-care preparation stage.

[0606] The device (200) may include available service data in end-of-life management information. The available service data is a service selected based on pet suitability data and condition suitability data among the target service data. For example, the device (200) may include actual available services in end-of-life management information among types of pet funeral preparation, types of pet transportation support, types of pet post-mortem care, types of memorial content creation, and types of memorial space usage.

[0607] The device (200) can include execution priorities for each available service data and scheduled execution times for each service in end-of-life management information. Through this, the end-of-life management information can be composed of schedule-based information indicating which services should be prepared first and when each service should be executed or guided.

[0608] The device (200) may include guardian input request conditions and guardian notification conditions in end-of-life management information. The guardian input request conditions may include items requiring additional input through the guardian's terminal (100), such as information on selecting a burial method, information on the transfer location, a desired schedule, a photo of the pet, a memorial message, or information on consent to use the service. The guardian notification conditions may include conditions for providing input request notifications, input deadline notifications, scheduled execution notifications, re-notifications, or service status change notifications.

[0609] For example, the device (200) can generate end-of-life management information including the execution priority and scheduled execution time of each service when the end-of-life management stage data is a preliminary preparation stage and the available service data includes funeral preparation services, transportation support services, and memorial content creation services. In this case, the device (200) can also include conditions for guardian input requests regarding funeral method selection information, transportation location information, pet photos, and memorial messages.

[0610] As another example, if the end-of-life care stage data corresponds to the emergency preparation stage and the level of need for end-of-life care is high, the device (200) can generate end-of-life care information with high execution priority for transportation support services and funeral preparation services. Additionally, the device (200) can include guardian notification conditions so that an input request notification or an execution schedule notification requiring immediate verification is provided to the guardian's terminal (100).

[0611] When generating end-of-life management information, the device (200) may include the basis for generating each data. For example, the device (200) may include in the end-of-life management information the reason why a specific service was selected as available service data, the reason why a high execution priority was calculated, or the reason why specific guardian input request conditions were set. This basis information may be used as guidance text displayed on the guardian's terminal (100) or as internal verification data.

[0612] The device (200) may include generated end-of-life management information in the service provision information. In this case, the service provision information may be provided to the guardian's terminal (100) in a form that includes not only the service name or simple guidance text, but also end-of-life management stages, available services, execution priority, scheduled execution time, guardian input request conditions, and guardian notification conditions.

[0613] The device (200) is not limited to providing end-of-life care information as medical diagnosis results or treatment plans. End-of-life care information may be platform processing information to support the use of end-of-life care services for pets, and may be information summarizing the availability of service provision, items requiring preparation, execution order, scheduled execution time, and guardian notification conditions.

[0614] Through this, the device (200) can provide services related to end-of-life care not as a simple recommendation list, but as actionable care information that reflects the pet's condition, service provision conditions, and guardian conditions. Accordingly, the service provision information provided to the guardian's terminal (100) may include specific information necessary for the selection, preparation, execution, and notification of end-of-life care services.

[0615] As a result, the device (200) can generate end-of-life care information including available services, execution priority, scheduled execution time, and notification conditions by synthesizing the level of end-of-life care required for the pet, the pet's condition, guardian conditions, and service provision conditions.

[0616] FIG. 6 is a flowchart illustrating the process of generating post-management information according to one embodiment.

[0617] Specifically, the device (200) can provide service provision information to the guardian's terminal (100), generate post-management information based on usage result data corresponding to the service provision information, and provide the generated post-management information to the guardian's terminal (100).

[0618] Post-mortem care information is management information generated by reflecting the guardian's usage status of the memorial service and healing content after the completion of the pet funeral or the commencement of the memorial service. The post-mortem care information may include at least one of memorial service suitability data, healing support suitability data, priority of provision by post-mortem care service, memorial content provision cycle, and healing content notification conditions.

[0619] The device (200) can collect post-mortem usage history data generated through the guardian's terminal (100) when information on the completion of pet funeral services or information on the start of memorial services is confirmed in the usage result data corresponding to the service provision information. The post-mortem usage history data may include at least one of an electronic memorial space access history, a memorial content viewing history, a memorial message input history, a guestbook input history, and a healing content usage history.

[0620] The device (200) can generate post-care information based on post-use history data and the guardian's preference conditions. For example, if the guardian repeatedly accesses the electronic memorial space and frequently views memorial content, the device (200) can adjust the memorial content provision cycle or generate post-care information corresponding to the type of virtual memorial content. Additionally, if the guardian allows the use of healing content and prefers meditation-based content, the device (200) can set notification conditions corresponding to the type of meditation-based healing content.

[0621] At this time, the post-care information is not limited to medical information for diagnosing or treating the guardian's psychological state. The post-care information may be platform processing information to support the guardian's memorial activities, content usage, and community linkage. Accordingly, the device (200) may provide memorial content, healing content, or community linkage guidance to the guardian's terminal (100), and may configure the post-care information in a form that does not presuppose medical treatment effects.

[0622] Through this, the device (200) can generate and provide post-care information based on the guardian's usage history even after the pet funeral or memorial service. Accordingly, the service provision information does not end at the time the funeral or memorial service begins, but can be expanded into post-care information that is continuously updated according to the guardian's memorial activities and content usage status.

[0623] Referring to FIG. 6, first, in step S601, the device (200) can extract pet funeral completion information and memorial service start information from the usage result data.

[0624] Pet funeral completion information is data indicating that a pet funeral-related service corresponding to service provision information has been completed. Pet funeral completion information may include at least one of the following: whether the funeral service has been completed, the date and time of completion, the funeral method, funeral facility information, whether transportation has been completed, whether the guardian has been verified, and funeral service identification information.

[0625] Memorial service initiation information is data indicating that a memorial service for a pet has started. The memorial service initiation information may include at least one of the following: whether an electronic memorial space has been created, whether memorial content has been registered, whether a mobile memorial space has been launched, whether a guestbook function has been enabled, whether a memorial message can be entered, whether virtual memorial content has been provided, and memorial service identification information.

[0626] The usage result data is data indicating the result of an actual service being applied for, reserved, executed, or completed in response to the service provision information provided by the device (200) to the guardian's terminal (100). The usage result data may be collected from at least one of the guardian's terminal (100), a service provision server, a funeral service processing system, a memorial content provision system, or an internal processing record of the platform.

[0627] The device (200) can distinguish between pet funeral completion information and memorial service initiation information by checking service identification information or service type information included in the usage result data. For example, if the service type of the usage result data corresponds to a pet funeral preparation type, a pet transportation support type, or a pet post-mortem care type, and the service status is indicated as completed, the device (200) can extract the data as pet funeral completion information.

[0628] The device (200) can extract the data as memorial service start information when the service type of the usage result data corresponds to an electronic memorial space type, a virtual memorial content type, or a memorial space usage type, and the service status is displayed as a start or active state. For example, when an electronic memorial space is created and becomes accessible from the guardian's terminal (100), the device (200) can extract the electronic memorial space creation information and the activation time as memorial service start information.

[0629] When the device (200) extracts pet funeral completion information and memorial service start information, it can verify the pet identifier, guardian identification information, service identification information, and processing time information together. Through this, the device (200) can distinguish whether the completion of funeral or the start of memorial service for a specific pet has been confirmed.

[0630] For example, the device (200) can extract pet funeral completion information by checking the pet identifier, funeral service identification information, funeral completion date and time, and guardian verification status from the usage result data. In addition, the device (200) can extract memorial service start information by checking the electronic memorial space identification information, memorial content registration date and time, and guestbook function activation status.

[0631] If some of the pet funeral completion information or memorial service start information is missing, the device (200) may classify the data as requiring verification rather than immediately confirming it as a target for post-management start. For example, if memorial content has been registered but the activation status of the electronic memorial space has not been confirmed, the device (200) may manage the memorial service start information in a pending state or receive additional verification information from the guardian's terminal (100) or the service provider server.

[0632] Through this, the device (200) can clearly verify from the usage result data whether the pet burial has been completed or whether the memorial service has been initiated before starting to generate post-care information. Accordingly, the device (200) can perform subsequent post-care information collection and generation based on the case where the actual completion of the burial or the initiation of the memorial service is confirmed.

[0633] In step S602, if at least one of the pet funeral completion information and memorial service start information satisfies a preset post-care start criterion, the device (200) may collect post-care usage history data including at least one of an electronic memorial space access history, a memorial content viewing history, a memorial message input history, a guestbook input history, and a healing content usage history from the guardian's terminal (100).

[0634] The criteria for initiating post-care management is a criterion set in advance to determine whether the device (200) will start generating post-care management information. The criteria for initiating post-care management may include at least one of whether the pet funeral has been completed, whether the time for the completion of the funeral has been reached, whether a memorial service has been initiated, whether an electronic memorial space has been activated, whether memorial content is available for provision, or whether the guardian has consented to the use of the post-care management service.

[0635] When the device (200) confirms that the funeral service completion status or the memorial service commencement status is confirmed in the usage result data, it can check whether the information corresponds to the pet identifier and guardian identification information. For example, if the funeral completion date and time is registered, or if an electronic memorial space is created and becomes accessible from the guardian's terminal (100), the device (200) can determine that the criteria for starting post-care management are satisfied.

[0636] Post-mortem usage history data is data indicating memorial-related usage acts or healing content usage acts performed by the guardian through the guardian's terminal (100) after the completion of pet funeral or the commencement of memorial service. The electronic memorial space access history may include the time of access, the number of accesses, or the duration of access, and the memorial content viewing history may include content identification information, the time of viewing, the number of viewings, or the duration of viewing.

[0637] Memorial message entry history and guestbook entry history may indicate records of guardians or authorized users entering messages or guestbooks in the electronic memorial space. Healing content usage history may include the date and time of use, number of uses, duration of use, or notification response status for art-based healing content, meditation-based healing content, counseling-linked content, or community-linked content.

[0638] The device (200) may reflect the guardian's setting conditions or consent status when collecting post-use history data. For example, if the guardian has not allowed the collection of community linkage or healing content usage history, the device (200) may not collect the item or may manage it in an inactive state.

[0639] For example, when an electronic memorial space is opened after the pet funeral completion information is confirmed, the device (200) can collect post-use history data including the number of times the electronic memorial space was accessed from the guardian's terminal (100), the time spent viewing memorial content, whether a memorial message was entered, and whether healing content was used.

[0640] Through this, the device (200) can obtain the guardian's usage behavior after the completion of the pet funeral or the commencement of the memorial service as basic data for generating post-management information. Accordingly, in subsequent processing, memorial service usage continuity data, healing content provision necessity data, memorial service suitability data, and healing support suitability data can be calculated based on actual usage history.

[0641] In step S603, the device (200) can generate post-use characteristic data from post-use history data, including access frequency, duration of use, number of times content is used, and whether healing content is used.

[0642] Post-mortem usage characteristic data is processed data generated to analyze how guardians utilize electronic memorial spaces, memorial content, or healing content after the completion of pet funeral services or the commencement of memorial services. Post-mortem usage characteristic data is not raw data that stores post-mortem usage history data as is, but rather feature values ​​organized to be used for calculating subsequent suitability and setting provision cycles.

[0643] The access frequency is data indicating the number of times or cycles in which the guardian's terminal (100) accesses the electronic memorial space or post-care service screen. The device (200) can calculate the number of accesses on a daily, weekly, or monthly basis, or calculate the average number of accesses over a specific period as the access frequency.

[0644] The duration of use is data representing the tim...

Claims

Claim 1 A method for providing an AI-based total life care platform service for pets, performed by a device, comprising: receiving pet registration data and guardian condition data indicating the guardian's service usage conditions from a guardian's terminal; collecting pet management data including health history data, biometric data, lifestyle record data, and service usage history data corresponding to the pet registration data; preprocessing the pet registration data and the pet management data according to preset preprocessing criteria to generate feature data for life cycle analysis; inputting the feature data for life cycle analysis into a pre-trained AI model to calculate pet life cycle status data and care need data; generating service matching data corresponding to at least one of a well-aging service, an end-of-life care service, a memorial service, and a healing content provision service based on the life cycle status data, the care need data, and the guardian condition data; providing service provision information corresponding to the service matching data to the guardian's terminal; and generating feedback data for AI model training based on response data to the service provision information received from the guardian's terminal and usage result data corresponding to the service provision information. The method includes the step of storing feedback data for training the AI ​​model in a pet life cycle learning database, wherein the AI ​​model is trained based on the pet life cycle learning database and is configured to generate life cycle status data and care need data as output data using feature data for life cycle analysis as input data, and the step of generating feedback data for training the AI ​​model includes service identification information, service selection status, etc., from the response data and the usage result data.A step of generating service usage characteristic data by extracting information on changes in the pet's condition at the time of service use and after service use; a step of matching the information on changes in the pet's condition after service use with the pet registration data, health history data, biometric data, and lifestyle record data based on a pre-assigned pet identifier; a step of generating pet-specific life cycle history data by sorting the matching results in the order of collection time; a step of excluding missing data, abnormal data, and data that does not satisfy observation period criteria from the life cycle history data according to pre-set exclusion criteria; a step of dividing the life cycle history data into reference interval data prior to the time of service use and observation interval data after the time of service use; a step of calculating changes in health status, changes in test results, and changes in lifestyle patterns by comparing the reference interval data and the observation interval data; a step of assigning life cycle labels according to pre-set life cycle label assignment criteria based on the pet registration data, health history data, biometric data, lifestyle record data, changes in health status, changes in test results, and changes in lifestyle patterns; and a step of comparing the changes in health status, changes in test results, and changes in lifestyle patterns with pre-set care effect judgment criteria. A method comprising: a step of assigning care result labels; a step of generating training candidate data by matching the service usage characteristic data, the life cycle label, and the care result label with each other; a step of classifying the training candidate data by combination of the life cycle label and the care result label; a step of correcting the sample ratio of the data by combination when the difference in data distribution by combination of the life cycle label and the care result label satisfies a preset imbalance criterion; and a step of generating feedback data for training the AI ​​model based on the training candidate data reflecting the result of correcting the sample ratio.AI-based total for pets Method of providing life care platform services. Claim 2 delete Claim 3 In claim 1, the method further comprises, after the step of calculating the life cycle status data and the care need data and before the step of generating the service matching data, a step of generating life cycle-specific measurement management information; and a step of setting improvement service candidate data included in the life cycle-specific measurement management information as service candidate data used to generate the service matching data according to a preset candidate setting criterion. The step of generating the life cycle-specific measurement management information includes: a step of extracting similar population data by comparing the pet registration data, the life cycle status data, and the health history data with individual history data stored in the pet life cycle learning database according to a preset similar individual selection criterion; a step of calculating impact data by measurement item by applying a preset impact analysis criterion to at least one of the activity level management need value, intake level management need value, bowel state management need value, sleep level management need value, and weight change management need value included in the care need data; a step of extracting items where the impact data by measurement item is greater than or equal to a preset impact criterion as core factor data; a step of generating precision measurement condition data including a measurement period, measurement method, and measurement precision conditions corresponding to the core factor data; and the precision measurement A step of collecting key factor measurement data corresponding to the key factor data according to condition data; a step of calculating measurement validity data by comparing the key factor measurement data with statistical reference data of a pet group belonging to the same species and same life cycle; a step of generating measurement deviation data and improvement item data by comparing the key factor measurement data with corresponding measurement data included in the similar population data; a step of generating improvement service candidate data based on the measurement validity data, the measurement deviation data, and the improvement item data; and the precision measurement condition data, the measurement validity data,A method for providing an AI-based total life care platform service for pets, comprising the step of generating life cycle-specific measurement management information including the above measurement deviation data, the above improvement-needed item data, and the above improvement service candidate data.

Images